First impression - its a decent pocket mythic to have, and comparatively better at lower stats. The upside is its kit makes it relatively easy to slot and at least be effective. The biggest downside is having basically nonexistent (some would say inverse) scaling.
I’ve tried it at 35 magic, and I’ve tried it at 6 magic. The 6 magic version does more burst damage (actually it fairly often got a three match off in the corner somewhere every other volley or so) while the 35 magic version was more consistent at clearing the board. However, since my attack stat is also pretty poor on the 6 magic version, I barely get any extra damage out of this, and still almost never an extra turn.
The troop does not have any significant synergy with Doomstorm. Doomstorm is, well, just kinda bad nearly all the way around because it doesn’t produce storm-like conditions for skyfall matches, which is the primary reason to use a storm. It does perform decently with a storm of its own color, or any storm it can be used to set up a chain with an ally troop. It does, for example, have a significant synergy with The Lord of Slaughter particularly under the effect of a Duststorm (eg,. Mountain Crusher, or just Runepreist if you can set up the initial kill somehow) - they share one (and only one) color, they feed each other very consistently with a high incidence of extra turns so long as you can get that initial setup and aren’t hard counterplayed. The 60 extra damage is not-insignificant when using this sort of setup, nor is the fact that the explosions on Ironhawk are not color restricted. It does, however, still suffer from losing a bit of potential mana for ever doomskull created, and on average before skyfall matches with 12 doomskulls created will only get just over about half its own mana back (generating about 3-6 mana for each color). I will concede that once you are dealing with health pools over 300, the damage barely makes a difference, and at this stage the 5 damage artillery support will in many situations do more harm than good (it is arguable that in a PvP setting, summon-on-damage is way more prevalent than mass barriers, the one thing it hard counters, and the trait is actually strongly detrimental in this sense.)
Not to harp too heavily on the spoiler data, we all know that wasn’t final, but using this troop for a while I can’t imagine what kind of situation someone was testing this at 20 doomskulls created and thought it was “overpowered” enough to have been scaled back. I can think of two edge cases - the “low magic cheese” where it could be used to 50/50 an extra turn massive doomskull strike or just a doomskull chained board clear if you “failed” and accidentally exploded one of the gems, and low stat concerns where you were concerned about the troop basically always annihilating the first troop in a low-stat scenario with a lot of easy setups for refills/extra turns. I think the way they scaled it back, if these two issues were indeed the issues being addressed and not actually the power cap of the troop (which I would quite frankly find ridiculous, because power-cap wise, under this design, it still doesn’t come close to top meta-troops)
If I had to make one change to try to fix this troop and maintain the current design, the change would have just been to swap the scaling value from gems exploded to doomskulls created. Change the explosions to a fixed amount (18 seems good enough to be “nearly the entire board”, and then change the doomskull creation to something like 6+ (magic /2). This would both weed out edge-case cheese where the doomskulls actually get matched (they are clearly meant to be exploded, not matched, otherwise magic would scale the other direction) and allow for better scaling - the caveat being that as damage scales up, primary explosion mana would scale down, but I wouldn’t consider this to be a strong downside as the troop is clearly meant to be used with storms of some kind which is where a lot of your mana is intended to come from. They have, in general, avoided troops that spawn gems that have their amounts scale with magic because they just become inifnites past a certain value and are completely unviable before a certain value, but I find it perfectly acceptable here since there is near zero chance any doomskulls would be matched under this design anyways, even less so when creating more doomskulls. So even if you reached 112+ magic and filled every tile with a doomskull, you’d deal 320 damage with a 100% board clear for zero explosion mana before skyfall matches. At the low end, this means it would be a slightly better generator while being worse damage at higher end deal more damage to the first target while being a slightly worse generator (than the current iteration) from the explosion that could still be set up with a storm. The mana cost could also stand to be 1 or 2 lower, but I don’t think that is the critical issue here.