Group Rollout Trainer Changes (#97)

* first changes

* core updates

* batch update

* fix typo

* missing import

* debug merge

* more fixes

* Remove dtype warnings

* Stub

* It runs

* Add in ref

* Pass linting?

* Remove extraneous 'calculations'

* Stub out push weights

* Remove tokenizer, add back in formatting

* Cleanup

* Updated default group

* update CI

* added tyro

* reverted build changes

Author: pbontrager