Ok, couple of issues you are going to have. Hooking in that many special unit types is likely to cause no end of trouble. The sheer amount of work that will come from just trying to make the tactics work with that many different kind of unit is going to be an utter nightmare. Further, there are much better ways to handle things. For instance, combat_pulse event triggers to enable a volley of magic, firing if a flank commander possesses a trait representing a unit of mages, using a variable to store the number of mages and thus the power of magic use, with an invisible combat pulse event having a chance to kill of your mages can be done for any major special units. You can also use such techniques to simulate artillery, which without the ability to expand combat, or to more reliably aggregate combat tactics, or add splash damage in the skirmish phase are likely to be far more effective in representing such powerful weapons.
Further, you are going to need a custom model for every unit, and with custom unit models you run into the limits of display. Properly set, you will either have special unit types always being the troop icon or not. Further you can only have one kind of special unit per holding. So many unit types rediculously expands the divergent building technology trees. Speaking of technology, you will now need to add new entries to the technology tree table to enable the proper 30-40% growth you can get from technology alone. Plus you will also need to add cultural and building perks that Vanilla unit types get. You have essentially multiplied the complexity of a combat engine that is explicitly incapable of representing complexity beyond a certain point. And this is going to be your nightmare.