feat: add livenessProbe to Package Manager

[stevenolen]

Summary

• Add HTTP /__ping__ livenessProbe to the Package Manager container, alongside the existing readinessProbe.
• Values match upstream PPM Helm chart defaults: initialDelay=10s, period=5s, timeout=2s, failureThreshold=10. Uses the named http port to stay consistent with the readinessProbe.

Test plan

• go build ./...
• Package Manager unit tests pass
• Deploy via adhoc image to a test cluster and confirm probe shows healthy

[claude]

Claude finished @stevenolen's task —— View job

---

Review Complete

• Read review guidelines
• Analyze diff and surrounding code
• Submit review

Submitted a non-blocking review. The change looks good — follows existing patterns and uses appropriate upstream defaults. Left two minor suggestions around field consistency and test coverage.

[claude]

LGTM — straightforward addition that mirrors the existing readinessProbe structure and uses sensible values from upstream defaults.

Minor observations:

1. Consistency with readinessProbe fields (package_manager.go:588-599): The readinessProbe explicitly sets SuccessThreshold: 1 and TerminationGracePeriodSeconds: nil. While those are defaults for liveness probes too, including them would keep the two definitions visually symmetric and make it clear nothing was accidentally omitted. Not a blocker.

2. No unit test coverage for the new probe: The existing package_manager_controller_test.go doesn't assert on probes. The flightdeck controller tests do verify their livenessProbe (flightdeck_test.go:197-200). Consider adding a similar assertion for PPM to catch accidental regressions. Also not a blocker for this PR given the test plan includes manual validation.

Overall: clean change, follows existing patterns, no correctness or security concerns.