fix(compressor): prevent orphan user turn after compaction (turn-pair preservation)

[teknium1]

Summary

Compaction no longer orphans the last user turn. When the last user message sits exactly at head_end (the first compressible index), the completed turn-pair is summarised as a unit instead of being split — so the summariser stops recording a finished ask as "pending" and the next session stops re-executing already-done work.

Root cause: _ensure_last_user_message_in_tail's final max(last_user_idx, head_end + 1) clamp returns head_end + 1 when last_user_idx == head_end, pushing the user past the cut into the compressed region without its assistant reply.

Salvages #22523 by @H2KFORGIVEN — reimplemented on current main (the original branch was ~6000 commits stale; the code moved from line ~1190 to ~2332).

Changes

• agent/context_compressor.py: add _find_turn_pair_end() and a Causal Coupling guard in _ensure_last_user_message_in_tail — when the clamp would orphan the user, push the cut forward to pair_end so (user → assistant → tool results) lands together in the summary.
• tests/agent/test_context_compressor.py: 8 new tests in TestTurnPairPreservation (4 for _find_turn_pair_end, 4 for the anchor incl. an end-to-end no-orphan invariant).
• scripts/release.py: AUTHOR_MAP entry for @H2KFORGIVEN.

Validation

E2E-verified on current main: the orphan reproduces exactly as described (returns head_end+1, user pushed out of the tail); the fix keeps the completed pair together and the tail starts on an assistant reply. #10896's mid-conversation pullback is unchanged (guard only fires when last_user_idx == head_end).

	Before	After
last user at head_end, cut>head_end	returns head_end+1 → user orphaned in summary	returns pair_end → user+reply summarised together
tail start after compaction	dangling user ask	assistant reply
compressor tests	125 pass	133 pass (8 new)

Infographic