Interrupting might be hard to implement, I think it is not something which goes well, with this turned based style.
The voice acting does not have to be, that "unique" in my opinion. Let say, that only priest/paladin, druid, wizard/sorcerer/warlock needs this.
That is 3 class only as spells are the same.
If voice action is only for group of spells maybe 30-35 different 6 sec voice line. (male/female as it was mentioned)
Making animation is harder, although I have no idea, what stage they are at, maybe it would take too much resource to change existing animations or to make them a bit more complicated or longer.
Regarding enemy spellcasting letting them animate is harder, without making fight longer:
If it is about enemy and team initiative, I would consider moving like 2 enemies at the same time (depending how much we face), to make combat a bit faster.
Same way you move 2 character when playing in multiplayer, with your friend.
That way you can have longer casting animation, when one warrior and one wizard moves the same time.
That is an other AI stuff, that might be also very-very complicated, and it might have compatibility issues.
