Late-game Town Enhancement.

At a speed of 8 (without tactics), there is a high probability that the "dances" will only affect one stack, which is not very fun. With tactics, it will almost always be at least 2.

With full armies, the probability of reaching only one stack without tactics is low.
No more than 0.5 STRICTLY (and 0.4 if you don't mess up). Taking into account the standard cover of archers of all types, it is about 0.2. And in what way is it high?

Is the ability to eliminate key stacks at the right moment not critical?

3 stacks of elves without tactics have a speed of 9, 2 of them are flying, and 2 more shoot. The elf will always reach the key stacks.
As for demons, tactics regularly HURT them against Might and Magic. An example from real life.
A demon with tactics moved its stacks, but the furies went first. The girls had an excellent choice of stacks, they could reach several, which means there was an opportunity to choose who to attack to deal with the elements. The horses were hit and a stack was immediately destroyed (due to the pit, they grew noticeably more). By the way, they would destroy a stack of dogs even without elements if they attacked in numbers. The dogs thinned out the Might and Magic troops, but there were still enough lizards left to carry away either the devils or the alt-lizards. Again, there is a choice of who to attack based on elements. In fact, they carried away the devils, and the lizardmen fought. The heroes hadn't gone yet, but the battle was essentially over, and even with a Might and Magic victory. I remember being very surprised at the time, it's a pity that replays have long been lost.
Hence, a simple conclusion: if the initiative is not in your favor, tactics can easily destroy the one who used them. And the one who didn't use them will win.
Only the elf can be sure of his initiative, and even then, with Winged Boots. Maybe a mage too.

The point is that in this match, one-shot kills happen constantly, and whoever strikes first wins. The furies go first, if the elemental doesn't work, then the orc is doing well, if it does and with luck, then it's very unpleasant.

The furies will have such a choice that you can consider the elemental to have worked for sure, and Luck with a probability of 0.5. (if there are no tactics, with tactics the probability of the elemental is almost always 7/16, but it can be even higher). At the start, the orc has very little Fury, if there is no Champion Artifact. The furies will destroy any stack except cyclops and, perhaps, leaders.
With luck, they can easily destroy the cyclops as well.

The archers' shot is weak, and the poison does not depend on the elemental.

No archers, only invisibility and we wait for our stack to cut them down with elements.

Next, the initiative is with the orc: shamans, cyclops (with the leader's boost) and warriors go before the lizards and dragons.

Yes, if the furies didn't destroy any of them. And if several possible elementals appear, the initiative is the first choice for attack. Sometimes the furies are even sacrificed if it blocks the desired stack and then allows the lizardmen to spread out comfortably. So, in the average scenario, let's say the shamans were destroyed, the leaders boosted the cyclops. And these may not even reach where they need to go. Why? Because units with a base initiative of 12 in the final game usually have an initiative of 14+ and go before the hero with Officer skill.
The warriors go at the same time as the lizards. If they stood next to the leaders, then again, they may not reach, because see above.
But if the lizards go intact, then the orc again loses a stack
with a very high probability. If it randomizes and destroys something with initiative, the dragons will also destroy their stack.
If you carefully sum up all the scenarios and all the probabilities, the outcome will be in favor of the orc, no need to consult a fortune teller. But when we are discussing the late game for Might and Magic, we should immediately forget about trying to come up with a mega-tactic to defeat everyone. At most, it is tactics and upgrades in which Might and Magic can win with some luck. In particular, the reasoning above shows quite real scenarios for victory, which will become even more real if you randomize the initiative (artifacts + starting distribution) and elements. Plus, you can periodically catch the orc on Darkness and Officer skill, making him always take the Darkness skill. If you manage to win 3-4 times out of 10, then the goal has been achieved, you shouldn't expect more from Might and Magic in the late game.

In terms of slots, it is easier for orcs to gain an initiative advantage through artifacts. For example, you can take the Necromancer's Staff for the weapon slot, but for the League, it is occupied by an axe, and then the League won't go.

I agree, that's why it's 3-4 out of 10.

In general, in the deep late game against the League without protection, you can not suppress anything at all, but take luck for reliability and go to kill, a la Winged Boots.
And there is also a cheater strategy with only cyclops, 2 stacks of shamans and 4 stacks of leaders.

This has been considered and tried many times :-) Officer skill + Darkness + staff + cloak of deadly shadows (Tactics and Readiness also help) = very good chances of winning against the orc without the Darkness skill for any race. With tactics and a boost, the expected number of hits from cyclops is something like 2 (too lazy to calculate exactly) and this is without blocking and reducing their initiative by fast creatures and with maximum morale. And this is an approximation with good artifacts for initiative for the barbarian.
Without the Darkness skill, these are not the scenarios that will please a strong orc in the late game. And there have been quite a few failures when the cyclops didn't even reach the enemy, but immediately went to kick their own.