feat(repo): carry the failure record to MCP, MSP, SARIF and traces

[systemfsoftware-maker]

Layer 5 (plan units U8-U11).

• MSP: an engine failure answers JSON-RPC error.data = the record, and the server keeps serving.
• MCP: rerun_mutant refuses with RerunEngineUnusable{code, record} through the typed error channel, and the new get_failure tool reads reports/mutation/failure.json.
• SARIF: a failed run with the sarif reporter writes invocations[0].executionSuccessful: false with one notification per record.
• Traces: tests pin stryker.failure.code on the run span and that the record's traceId equals the exported trace id.
• Log noise: the "Possible causes" block appears only when an option is unknown, and the bundled parser's WASI ExperimentalWarning is dropped.

Residuals

• P3 (review, not applied): mcp-server.cell.ts engineRefusalOf and Serve.cell.ts die with a defect when an engine failure classifies as something other than RunFailed. Today no engine path can produce that (the conclusion classifier only yields RunOk for help or success exits), so the die guards an impossible state. Convert it to an InvariantBroken record if that changes.

[ryanleecode]

Reviewed by the merge conductor as part of the diagnostics stack. CI green on this head except where noted in the stack summary (macOS timeouts on #151 are flaky: the same code is green on macOS in #155 and #156; the #156 stryker-js mutation shard times out until main records its first full-cost timing, as this PR's description says). Merging into feat/agent-ready-failure-diagnostics with a merge commit.