docs: forced-subtag fix resolution (#1 proto artifact, #2 fixed)

docs/plans/2026-06-15-forced-subtag-mode-fix.md

@@ -95,3 +95,22 @@ BuildConsumer→SubtagAlarmConsumer, or a contract/IPC round-trip test if the en
 Branch off `main`. #2 is the clean confirmed fix; #1 is diagnose-then-fix. Net48 worker
 constraints apply (no init-only props/positional records). Do NOT increment `provider_switches`
 on an initial forced-mode set.
+
+---
+
+## Resolution (2026-06-15)
+
+**#1 was NOT a bug — it was a grpcurl proto-mismatch artifact.** End-to-end instrumentation
+proved: the gateway sends `forcedMode=Subtag`, the worker's `BuildConsumer` builds the
+`SubtagAlarmConsumer`, and the worker `QueryActive` + gateway `ApplyReconcile`/`StreamAlarms`
+all carry `degraded=true` / `source_provider=SUBTAG`. The original "degraded=false" observation
+came from running grpcurl against the host checkout's proto (branch `feat/lazy-browse-children`),
+which predates the feature and lacks the `degraded`/`source_provider` fields — grpcurl silently
+dropped them. With the matching proto, every active alarm shows `degraded=true`. No code change.
+Tasks 1/3/4 (worker diagnostics + #1 fix) were dropped; the temporary diagnostics were reverted.
+
+**#2 was a real defect — fixed.** Gateway now seeds `_providerMode`/gauge/feed from the configured
+forced mode (`fix: gateway reflects configured forced provider mode`). Verified live:
+`provider_mode=2`, first `ProviderStatus` = `Mode=Subtag, degraded=true,
+reason="Forced subtag mode (configuration)"`. Auto mode unchanged → windev production (Auto)
+unaffected; no redeploy required. Gateway tests: 163 passed.