fix(session): per-turn history check is O(1), not O(n) per turn

[alpibrupa]

Summary

• run_turn_(streaming_)with_provider re-derived the ENTIRE conversation history from the trail log and content-compared it against the in-memory cache on every single turn, to catch a failed evs.record_assistant append (finish_turn's own comment: "the cache keeps the message anyway"). That re-derivation is O(total history) per turn, so a session that runs long enough is O(n²) overall.
• Reproduced live: a synthetic session doing nothing more than ordinary turns of realistic-length assistant text crashed with a step-limit-exceeded panic in lex-schema's json_value parser by turn 46, purely from re-parsing an ever-growing history on every turn — independent of any one message being unusually large. This is what was actually blocking a real multi-file package build via lex-code's agent loop.
• Replaces the full re-derivation with evs.event_count: a cheap SELECT COUNT(*), compared against the in-memory cache's length. A failed record_assistant append shows up as a count mismatch exactly as reliably as a full content comparison would, since the trail is append-only. expected (the cache plus this turn's input) stands in for the old derived once the count agrees, since that is what a full derivation would reconstruct anyway absent a divergence.
• Trade-off, stated plainly in the code: this no longer catches content that silently changed underneath without changing the row count (e.g. the non-ASCII-collapse scenario fix(session): surface a refused turn to the user instead of silence #135 was about) — only that nothing has gone missing. Accepted because the check it replaces could crash the whole process outright on a long session, which is worse. fix(session): surface a refused turn to the user instead of silence #135's on_step-routing fix stays needed regardless, for the failure modes event_count still does catch.
• session_history itself is unchanged and still used for its full, content-verifying cost paid once at session resumption (resume_session) rather than per turn, and its own tests are untouched.

Depends on alpibrusl/lex-schema#36 (merged today), which fixed a compounding quadratic bug in the same code path (escape-dense string parsing) that could still panic within a single turn on verbose model output, independent of this fix.

Test plan

• lex check on both modified files, plus a full repo sweep (lex check over every tracked src/ file)
• lex fmt --check src/
• lex test (4/4 passing, including test_session_events.lex which still exercises session_history directly)
• lex doc-sync --check
• A standalone probe reproducing the exact crash: 46 turns of ordinary-length assistant text hit the step-limit panic with the old per-turn full derivation; 200 turns of the identical workload complete cleanly with event_count
• End-to-end: re-ran the real multi-file package build that originally surfaced this (an agent implementing an RLP codec) against a fresh project with both this fix and lex-schema#36 in place

🤖 Generated with Claude Code