Tokens (inputTokens, outputTokens) now come from ACP PromptResponse.usage
which is populated synchronously from run_conversation() — no WAL race.
Turns still come from DB before/after snapshot.
Previously both were read from hermes state.db after ACP prompt returned,
but WAL write lag caused incomplete token data (e.g. 235 vs actual 26,080).
Refs #91
On session resume, turns/inputTokens/outputTokens were cumulative
(entire session history) instead of per-step increments. Now we
snapshot metrics before prompt, compare after, and report the delta.
Changes:
- acp-client: add getSessionId() accessor
- hermes: extract snapshotUsage() + computeUsageDelta() pure functions
- hermes: runPrompt/runHermes/continueHermes use before/after snapshots
- 9 new unit tests for usage delta computation
Refs #68
xiaoju