fix: handle mixed batches with missing saved modules

[rushikeshraghatate90]

Summary

• Fix mixed adapter batch forwarding when an adapter does not have modules_to_save for a wrapped module.
• Fall back to the original module when the active adapter is not registered with the auxiliary training wrapper.
• Add a regression test covering mixed batches containing two adapters and base-model inputs.

Testing

• python -m pytest tests/test_custom_models.py -k missing_modules_to_save -vv --no-cov — passed.
• python -m pytest tests/test_custom_models.py -k TestMixedAdapterBatches -q --no-cov — 35 passed, 2 skipped.
• The full test file was interrupted before completion.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[rushikeshraghatate90]

Closing this PR to avoid duplicating the original contributor's work. Apologies for the oversight, and thanks for the clarification. I'll leave the issue to the original contributor to submit their fix following maintainer approval.