fix(canon): correct SSE bytes_out claim in telemetry-validation-gate

[klappy]

Disposition of Cursor Bugbot finding on PR #210 plus a related factual error caught by smoke evidence.

What Bugbot caught on #210

Bugbot flagged the original procedure step 3 as referring to request_body and response_body (full HTTP bodies including JSON-RPC framing) when the wrapper actually measures JSON.stringify(args) and JSON.stringify(result.content). That finding was addressed in the merged version of #210 before merge.

What this PR fixes

The merged version still contains an SSE claim that contradicts the wrapper's design:

For SSE-streamed responses, expected bytes_out = 0 and tokens_out = 0 per the Emission Contract.

The wrapper measures the in-memory content envelope before MCP transport framing. SSE framing is applied later, on the wire. The wrapper sees the JS object, not the SSE-encoded body. Therefore SSE-transported responses produce non-zero bytes_out at the wrapper.

Smoke evidence (main preview, 2026-05-16 00:30Z)

All 14 successful SSE-framed tool calls in the smoke pass emitted non-zero values:

tool	bytes_out	tokens_out
oddkit	671	235
oddkit_audit	16,282	4,679
oddkit_catalog	28,593	7,780
oddkit_challenge	20,638	5,221
telemetry_policy	28,014	6,681
... (and 9 others, all non-zero)

Zero SSE responses produced zero emitted bytes_out. The wrapper's in-memory measurement is exactly the failure mode the Emission Contract was written to defeat.

The fix

Replace the wrong SSE caveat with a correct one that names the old wire-edge instrumentation as the source of the "0 for SSE" line in telemetry-governance §Numeric Values.

---

Note

Low Risk
Low risk documentation-only change; clarifies expected telemetry calculations for SSE-streamed tool responses to avoid incorrect validation outcomes.

Overview
Updates canon/constraints/telemetry-validation-gate.md to remove the claim that SSE-streamed tool responses should emit bytes_out = 0 / tokens_out = 0 during validation.

The gate now explicitly states that the wrapper measures the in-memory { content: [...] } envelope before transport framing, and that the “0 for SSE” caveat applies to legacy wire-edge instrumentation, not the current wrapper.

^{Reviewed by Cursor Bugbot for commit 81d58d8. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

Canon Quality — oddkit_audit ✅

No dead klappy:// references or legacy link patterns found in writings/. 42 files scanned.

_{Spec: klappy://docs/oddkit/specs/oddkit-audit · Workflow: .github/workflows/canon-quality.yml · Run: #155}

[github-actions]

Canon Quality — Frontmatter Schema ✅

All 41 file(s) in writings/ conform to klappy://canon/meta/frontmatter-schema.

_{Validator: scripts/validate-frontmatter.py · Canon: klappy://canon/constraints/frontmatter-validation-before-merge · Run: #155}