Add truncated importance sampling and DrGRPO args

[jacklanchantin]

What does this PR do? Please describe:

• Adds tis_imp_ratio_cap to use truncated importance sampling correction
• Adds GrpoLossConfig adv_std_normarlization (for DrGRPO)
• Adds new if statement to skip ref_logps computation for kl if beta == 0 (as done in DrGRPO)
• Adds GrpoLossConfig loss_token_mean for normalizing over all tokens

Fixes #{issue number}

Most importantly, this adds truncated importance sampling correction, as recommended by @uralik.

Does your PR introduce any breaking changes? If yes, please list them:
List of all backwards-incompatible changes.

Check list:

• Was the content of this PR discussed and approved via a GitHub issue? (no need for typos or documentation improvements)
• Did you read the contributor guideline?
• Did you make sure that your PR does only one thing instead of bundling different changes together?
• Did you make sure to update the documentation with your changes? (if necessary)
• Did you write any new necessary tests?
• Did you verify new and existing tests pass locally with your changes?
• Did you update the CHANGELOG? (no need for typos, documentation, or minor internal changes)

[jacklanchantin]

only use kl if ref_logps were computed

[swarnaHub]

nitpick: this also means that beta is non-zero? does an assert make sense that it should never come here if beta is zero? or something that makes this if statement conditioned on beta for better readability?

otherwise LGTM!

[uralik]

lets make sure this works for bs>1 before merging ! (as discussed offline)