fix(flows): drop unpairable trailing FRs in rearrange

[a2105z]

Summary

Follow-up to the orphan-FR prune that landed in 73e8625 / discussion on #6587.

_drop_orphaned_function_responses only removes responses that carry an id. Trailing id-less / empty-id FRs, and rearrange itself when the prune is bypassed, could still raise during contents assembly and permanently poison the session.

This change:

• Makes _rearrange_events_for_latest_function_response drop unpairable trailing FRs instead of raising (defense in depth requested on fix(flows): Drop orphaned FRs instead of poisoning sessions #6587)
• Strips unmatched FR parts when a call event is found but the trailing event also carries extras (same poisoning ValueError)
• Leaves true id-less FC/FR adjacent pairs alone (model families that strip ids)

Fixes #6751
Related: #6582, #6587

Test plan

• uv run pytest tests/unittests/flows/llm_flows/test_contents_function.py tests/unittests/flows/llm_flows/test_contents.py -q → 56 passed
• Repro cases from contents rearrange still raises on id-less / unmatched trailing function responses (follow-up to orphan FR prune) #6751 (id=None, id="", multi trailing id-less, rearrange bypass, unmatched extras) no longer raise

[a2105z]

Hi @GWeale @DeanChensj @llalitkumarrr — this is the follow-up from #6587.

After 73e8625 pruned id'd orphans, rearrange could still raise on id-less/empty-id trailing FRs (and on unmatched extras). This makes rearrange drop those instead of poisoning the session, as you suggested.

Ready for review whenever you have a chance — thank you!