I think the critical path for getting the game out is primarily artistic. You need the story and its potential dialog paths to be solid, the art assets ( environment/props/creatures ) created, along with animation/voice acting and soundscape/soundtrack creation, before you have any hope of a playable game.
Sequencing the production/acquisition of these assets determines the minimum time to release, as they are needed to be able to create levels/scenes, as well as script and test encounters and activities that comprise the game.
Certainly the game engine needs to be reasonably solid, with minimal crashes and functioning systems before you can put out EA, but refining the engine behaviour, the game appearance, and individual area balance are probably weighted to later in the cycle, particularly as there systems still being built to support classes/races/spells/creatures that are not yet available.
I would certainly prefer that Larian manage their resources to most efficiently complete the game, to minimise the likelihood that content is cut or story quality/options diminished. If this means fixing animations and other options, changes and tweaks are pushed back to later in the EA, then that is fine as far as I am concerned.
