IMO, items from games of this genre are generally not well thought out. Devs do not design skills and items so that each one of them could fit into a certain playstyle or build. Instead, they just design items randomly and expect players to try out and develop their own build. When players played sufficient hours and created sufficient amount of different builds, devs then balance the stats on items or redesign some of them, or add more items or dlc based on players' feedback and data collected. This could apply to Diablo II and Path of Exile, and I think it also applies to GrimDawn.
However, in this game, the original devs never had a chance to wait that long to make such changes. The current devs...well I'll just skip my comments here.

OK I tested, and confirmed that block happens BEFORE resistance take into effect.
The test is as follows. Find a shield, in my case I bought one from Epic A1, which has 193 block amount. Go to Legendary A5, find these Empusa, kill all but leave one alone. Now unequip everything that grants the resistance of that Empusa does, and see how block works. No block at all.
Now equip everything that grants resistance of that damage type, make it as high as possible. In my case I made my fire resistance 69%, and the damage from Empusa per hit is around 140, which is lower than the shield block amount. Still, as I expected, still no block at all.
If resistances is to work before block, then I should be able to block the attacks from Empusa after I raised my resistance.
The test above is very easy to repeat. You can test it on your own if you don't believe in that.

Note:
Consider the test above the damage and block amount is kinda close, I did a further test. This time find another shield with 329 block amount and further raised my resistance so that the damage is around 100 per hit. Still no block.

titanquest.wikia also says it happens after resists

http://titanquest.wikia.com/wiki/Shields

Quote

Example

A projectile gets blocked by a shield with a Chance to Block 60 Damage.

If not blocked, after resists that attack would do:

50 Fire Damage from weapon (50 < 60 → blocked, 0 damage)
60 Fire Damage (60 = 60 → blocked, 0 damage) and 70 Vitality Damage (70 > 60 → not blocked, 70 damage) from an artifact
111 Burn Damage over 3.0 Seconds from a skill (111/3 < 60 → blocked, 0 damage)

it specifically said after resists

there is also this line

Quote

The damage is already reduced by armor and resistances.

That's interesting. Everything is exactly what I said except this one. Let me think about how to test it.

I have tried to convince other players in other forums that quick recovery is overrated in current version, but few agreed. So it's glad to see someone who tried and holds the same opinion.

So before we say something is good or bad, I think we should look at the mechanism first, and that is where the problems raise.
The first mechanism is what botebote77 already mentioned, block recovery cap: In AE/R, shield recovery time (or maybe just call it block cooldown) has been capped at 90%, which is 0.25 sec or 0.3 sec.
Personally I think that is the second most stupid change AE made. Obviously the most stupid change is the huge nerf in scatter shot which also has been mentioned above. I am not saying scatter shot is not OP, it deserve some nerf but defenitely not reduce its damage by 80%. To those who think the bleeding damage would compensate, I would like to say that the bleeding damage from scattor shot now does NOT stack with bleeding damage from any other sources. So even undead isn't a problem, this skill is completely ruined. Well I'm a little bit off topic...
What does the 0.25/0.3 sec mean? First, when you face a pack of monsters, with the cooldown in block, you cannot block all of them as you did in IT. Instead you can only block some of them. The higher the number of monsters and attack speed of monsters, the lower % of attacks you can block. Or in other words, the more dangerous situation you face, the less block will help. Second, some boss and hero skills are actually multi-projectiles, although they may not look like so. Without the cooldown cap in block, you can block all of them which means you take literally zero damage. However with that stupid coolddown, you can only block one for these projectiles, which means you take almost full damage.
Based on the decription above it might seems block is OP in IT and should be nerfed. But there is another mechanism that prevented block to become actually OP, and that same mechanism has made block even less useful in current version, which is: you can only block damage that lower than the shield block cap. Any damage that is higher than the damage a shield can block, will simply bypass block and do full damage on you. According to what I learned, that "damage" refers to the number which already calculated all the damage modifiers from attacker's side (in this case monster's), but not defender's resistances, armor and damage absorption. For DoTs, use the "damage per second" number to compare, not total damage. So for example, an attack that would deal 800 damage to you but actually deals 400 because of 50% resistance, a shield with 799 block amount cannot block it. In reality, with good shield in accordance with your progression equipped, it is not so common in IT to find some monster damage that would bypass shield block cap. However, such situation becomes pretty common in AE, since the increase in damage of monsters is higher than the increase in shield block cap, and new purple shields are not so satisfying in other stats so they are generally not used in the first place. So generally speaking, it's good agains monster with lower damage, but do nothing to stronger ones.
To make things worse (or better in certain situations), the system will first roll if an attack is blocked or not. If yes, then check every single damage of that attack, see whether it is higher than shield block cap. Regardless of the result, block is now in cooldown. So it is possible that you blocked, but all damage bypass block cap, nothing happens but your block is in cooldown. It is also possible that a seemingly very high damage attack which consists of several small damage can be fully blocked.
Moreover, for melee attacks, block is checked after crit check. So if it crit, it's the crit damage that wuold compare with your shield block cap. And since the actual block CHANCE is related to DA/OA, it is much harder for a character with low DA to block melee since first you have less chance to block and even if you block the crit damage is very likely to bypass it. On the other hand, a character with high DA is much easier to block due to more chance to block and the reverse-crit. Now that go back to the similar conclusion that has already occured twice: when you are less reliant on block (you have high DA), block performs well; when you really want block to help you (you have low DA), block performs poor.

To conclude, the whole block mechanism works better on smaller monster pack, lower enemy attack speed, less damage of single attack, and higher DA characters; and work poorly or totally doesn't work on opposite situation. That says, block will makes you safer when you are already not in danger, but it will not likely to save you (although sometimes it will) when you face real challenge.

I am the one who voted int based. By int base, I mean put some points in dex (usually up to 425 to equip Stonebinder), and the rest in int. Invest as little str as possible.
Since OP already mentioned pros and cons for str/int based Runesmith, I'll only answer several questions raised above. To clearify, I did the theory-craft but my friend actually played it, not myself. So I may have some mistakes in details.

1. Where does the base damage come from?
Transmutation of course. It will transfer 99% (at max+4 level) of your weapon and shield damage to elemental, but the added flat damage in Defense mastery skills will not be transfered as bote already mentioned. The BIS weapon for elemental Warfare/Defense, IMO, is an axe, Ritual Cleaver. Its accumulated base damage is more than 400, if I remember correctly. That also partially solves attack speed problem. Resistance reduction is not in very high priority for this build since you are dealing elemental damage, which is the combination of fire, cold and lightning. Only a few monsters have very high elemental resistances and if they do, you still have the Mega Explosion to deal with them. So I would personally skip Sapros, it's really not mandatory and the reduced max life is not easy to compensate with this build.

2. Is it required to equip str-based armor?
Not necessary, I don't think there is any str armor that you can't miss. But if you wish, you can do that. Armor handling from Defense is not sufficient to equip endgame str amor alone. So at least you need one additional item to reduce gear requirement, or the invested points in str would be too much. One good choice for instance, Odysseus' Armor. Alke and Timaios is not a good endgame choice with the high opportunity cost, but they are decent.

3. Warfare and Defense
I don't think Defense is inferior in terms of int based melee. The only advantage for Warfare is that endgame weapon has much higher damage than shield (as I mentioned above). OA and attack speed can be compensated with good gear. Defense has its own pros. The resistances provided by Defiance and Iron Will are huge, and shield charge gives you advantage over ranged monsters which warwind cannot.

So my conclusion is that this build is pretty viable and not handicapped at all. Yes it is more gear reliant than str based, but when you get these gear, it has higher potential.

Ha, the infamous tt300.
Never expect anyone mentioned this guy here.

Long story short.
You need
https://tq-db.net/equipment/golden-agris-helmet
and
https://tq-db.net/equipment/gambanteinn
Note that both items have 50% phy-to-ele. But due to unknown reason it is not shown in the database. But trust me they do have such stats.

TQ               :  (pierceDamageDV * ((dexterityDV/650)+1))+(dexterityDV*0.03)

TQ-IT 1.17 :  (pierceDamageDV * ((dexterityDV/1000)+1))+(dexterityDV*0.03)

TQ-AE         :  (((pierceDamageDV / ((strengthDV/500)+1)) * ((strengthDV/2000)+1)) * ((dexterityDV/400)+1)) + (dexterityDV*0.03)

You somewhat...confused by two types of pierce damage: flat pierce damage and pierce damage converted from weapon.
Flat pierce damage does not benefit from any attribute pointes in TQIT, and does benefit from dexterity in AE (1000/100%).
Pierce damage converted from weapon benefit from both strength and dexterity, both in TQIT and AE. The difference is that in TQIT, the multiplier is 500/100% for strength and 1000/100% for dexterity, and 1000/100% for strength and 500/100% for dexterity in AE.