fix: remove drain-state operational fingerprint from /healthz endpoint (PILOT-335)

[matthew-pilot]

Summary

PILOT-335: /healthz endpoint exposed drain state to external observers (operational fingerprint).

Root Cause

beacon/server.go:1228-1234 — ServeHealth() returned different HTTP status codes (200 vs 503) and body text ("ok" vs "unhealthy") depending on internal drain state, allowing an external observer to detect when a beacon enters graceful scale-down.

Fix

• server.go: /healthz always returns 200 "ok" regardless of healthOk state. SetHealthy() is preserved for internal state tracking.
• server_test.go: Updated TestHealthEndpoint to verify consistent 200 response after SetHealthy(false).

Verification

• go build ./... ✅
• go vet ./... ✅
• go test -run TestHealthEndpoint ✅

Size

2 files, +7/−11 | matthew-fix (small tier)

🔗 https://vulturelabs.atlassian.net/browse/PILOT-335

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[matthew-pilot]

🤖 Matthew Pilot — CI Status

Check	Status
test	✅ pass (1m8s)
codecov/patch	✅ pass
Mergeable	✅ MERGEABLE

Verdict: All checks green. Ready for review.

[matthew-pilot]

🤖 Matthew Pilot — PR Summary

PILOT-335 — Fixes an operational fingerprint in the /healthz endpoint.

What changed

beacon/server.go — ServeHealth() now always returns 200 OK regardless of whether the server is draining. Previously it returned 503 when unhealthy, which let external observers detect graceful scale-down timing.

Root cause

server.go:1228-1234 leaked internal drain state through HTTP status codes and body text ("ok" vs "unhealthy").

Changes

File	Δ
server.go	+2/−6 — Always 200 + "ok"
server_test.go	+5/−5 — Updated test expectations

Verdict

✅ Builds, vets, and tests pass. 2 files, +7/−11, small tier.