Broken evaluate function

The evaluate function doesn't redistribute randomly halite between plays. I suspect it may concern other games too. I also suspect it comes from the fact that in core.py self.done is never set but its value is used everywhere to determine wether to reset a game or not. Games are still reset though (but with same initial distribution) - maybe some internal mechanisms of each environment.

Start by tracing the evaluate function and the self.done handling in core.py, then compare how game resets and initial halite distribution behave across the affected environments. Reproduce the issue with repeated evaluations and verify that games reset with randomized halite before considering the work complete.

Written by the indexing model from the issue text.