I imagine the philosophy will still be "no cutscenes outside of quests and no tutorials after early game." Can fast travel be immersion-breaking? Sure, but it doesn't have to be. The same goes with interacting with NPCs, anything that they did would be far too annoying after the first time.
I sympathise with the sentiment: it's true that sixy little animations can be fun the first few times, but then quickly get annoying. Like opening a map, for example. It might be cool the first few times to watch your character take out a map from his pocket (or wherever) and unfold it. But it'll quickly get annoying, because you just want to bypass it to look at the map.
But I think there are (at least) a couple of differences with spell training. Firstly, we know that there are only about 80 or 90 spells (I may not be recalling the exact number correctly, but I'm pretty sure that's roughly right). So the spell training animation won't be seen all that regularly, and it has an upper limit. Secondly, I think there's the potential here to make the spell training animation interesting and useful: there could be a different animation for each spell, where the seller
demonstrates the spell to you. One might it imagine the situation going like this. The seller says "you say the magic words, and this is what happens" *zap*.
I mean it could be something as simple as this (lets say you go to a spell merchant to buy a fireball spell):
The merchant takes you out back to some kind of target, or possibly they conjure a low level enemy. Then they say "watch closely" and cast the fireball. Then the spell appears in your hand, and you practice it a few times on the conjured enemies until you tell the trainer that you're done.
That's all I would need to feel immersed. The trainer casts a spell once or twice, then instructs you to do the same. Maybe your spell could initially fail, but after a few attempts you understand it and can use it properly.
Yep, I think this would work nicely.