2 Vbn:
I hope you're not offended by my attempt to rush you.
The description turned out to be very clear and instructional overall. Thank you for your hard work.
Regarding the details:

We have 27 snipers, which means the expert Archery skill is necessary. The first person (after gaining 1 level) who can obtain it is only Enova, since she has advanced Archery at the start. Therefore, the choice was not difficult, especially considering that Enova was given Luck.

An interesting point that is important to me. Why is betting on expert Archery more important than betting on expert Luck (Leadership)? At first glance, rolling several L or Morale checks more than offsets the additional damage gained from expert Archery. Or am I missing something?
And in general, was an option considered, for example, with Myllich (expert Leadership, starting Archery at level 3)? And if so, why was it rejected?
Of course, generating Morale specifically for snipers in the console is very tedious, but perhaps with Myllich it would have been possible to get Slow from the wizard on the first turn?
Besides, the battle with the titans performed by Myllich would have made it possible to save the angel. After all, the wizard could cast "Haste" BEFORE the titans' turn, and then a harpy could be used to block them instead of the angel. Minus 1 to morale, of course, but this penalty disappears after the harpy dies. Or does it?

Now for the step-by-step description.
Day 1
05) We buy Enova (followed by the Dragon mage).
Day 1
18) We buy Pir (followed by the Dragon mage).

After Pir, was Moner likely put into the tavern, and not Dragon?

For example, it was not possible to achieve the Wizard casting Slow in the 1st round.

Meaning, the battle map was stable, but none of the initial troop layouts by slot provided an option with the wizard having Slow on the first turn. Correct?

16) Kilgor takes the artifacts and troops from Gretchin, the latter resigns, and Kilgor destroys Haart, losing a harpy and a couple of troglodytes, then visits the Cyclops Vault

This was the only dark spot for me in your walkthrough and, as luck would have it, you only mentioned it in passing. :)
Kilgor is good, and together with Rumble - even better. But as for the troops - only 5 harpies, 4 snipers, and trogs. Did you also take a wizard and generate Slow? Or how? Please describe this battle and the losses.

26) It should be noted that with 3 markets in the mercenaries guild, it was more profitable to exchange some troops for gold, and some for resources.

Most likely, there is a base price for all creatures in gold. When converting to gems, the result is rounded to the nearest whole number. I should take note of this nuance.

Many people here talk about generating morale using delays. If this is news to anyone else, let me inform you: there is an authoritative opinion that morale (luck?) can be manipulated by performing certain steps in battle after a specific interval of time. But it is still unclear what these delays depend on and what they equal for each specific case.
But the question is quite interesting, and I would, of course, like to study and discuss it.

From what has been said, at least three interesting conclusions can be drawn:
1. The calculation algorithm for rolling L (Morale) is launched not after the mouse click, but BEFORE!
2. During the pause (from the moment the turn begins to the moment of the mouse click on the action), the game's program code repeatedly, with some periodicity, runs the calculation algorithm, giving a different result ("Yes" or "No") each time. And this result is valid until the next launch of the algorithm.
3. One can even roughly estimate the periodicity of these launches. From 0.5 to 2 seconds. Because if the algorithm were launched every 0.1 sec, we would not be able to notice or track it, and any simple counting rhyme would have an error greater than the duration of the period. For similar reasons, the period is unlikely to be 2 sec or more.

And I started writing about all this because I actually wanted to share another observation. When conducting the battle with the titans on one computer (Windows XP) and another (Windows 2000), the course of the battles was different, and it was impossible to achieve on the 2nd computer what happened on the 1st, and vice versa. But for the console, everything repeated regardless of which computer the battle was conducted on.

It looks very much like the issue is not the version of Windows, but the personal characteristics of each specific OS. So to speak, in the unique individual fingerprint. Including all installed applications, the configuration of the specific computer, and even the "virus bouquet" :) . This version is supported by the fact that similar suspicions existed when studying randomness in other games, not HoMM. Additionally, when reproducing the battle in the console from your save, my wizard generated different casts. With the same sniper actions (always pressing "Shield"), the wizard generated:
1. Bloodlust.
2,3,4,5 times: Haste
6 times: Slow! :smile03: :)