A program for determining luck/morality in Heroes 3!

If suffering isn't an argument, then what's the difference between +1 Block and +3?

Well, you're going to extremes. :)
If the suffering of Enova's snipers and the slightly greater suffering of Mullich's are comparable, and both cases require hundreds of reloads, then suffering isn't an argument.
But, for example, with Titans, I needed Block to drop 5 times in a 7-round battle. With a +1 Block skill, it could have required tens of thousands of reloads and 2 weeks of continuous loading. Clearly, there's a difference here, and I might not have achieved the result even with suffering.
I'm not a great expert on battles, not even a small one, so I'm asking more experienced players: how realistic and how preferable is it to:
have +3 Archery and 1 Block, or with three Blocks, can you create something so powerful that Archery isn't really needed.
It would also be good to figure out if there's a difference between open battles and battles "indoors," in terms of priorities between Archery and Block.

But Archery does make a difference! As we can see from the battle video, Enova's snipers kill 17 griffons with a single shot, meaning that with two hits, the entire stack is guaranteed to die. And how many would Mullich kill with basic Archery? Well, 12, maybe 14 at most, and that's clearly not enough to kill the entire stack in 2 hits...

I might be saying something stupid, but I read somewhere about a tactic based precisely on NOT finishing off stacks of griffons. While these stacks of a few units are attacking the sniper screen, they are shooting down full, large stacks that have nowhere to fly in yet. Is there such a tactic, or is it nonsense?

P.S. I watched the video about Titans, but the link to the forum post doesn't open.
P.P.S. I managed to test this delay a little. It's already clear that this hypothesis has some basis. Here's the data.
In the champion battle with Titans, I did it like this: if Block didn't drop for the snipers in the first turn, I immediately reloaded. It dropped roughly 1 out of 10-15 reloads.
Now I tried experimenting with the delay - the results immediately changed dramatically, on average I managed 6-8 times out of 10 attempts. But I have the impression that the delay time needs to be felt much more subtly than I thought. That is, if there is indeed this periodicity of random change, then it lies within the range of 0.3 - 0.5 seconds. If you don't hit exactly this interval, Block is lost. But, of course, more testing is needed.