Improving integration mode to become more resilient to upstream outages

[bennyyang11]

What does this PR do?

Shortcut 136332

Makes the SDK pod survive brief replicated.app outages.

• Bootstrap is split: the listener comes up on local-only work; upstream sync runs in the background and no longer blocks /healthz.
• Integration mode caches the license to a replicated-license-cache Secret and serves from cache when upstream is down (sets X-Replicated-License-Cache: stale).
• Opt-out via requireUpstreamOnStartup: true in chart values.

Does this PR introduce a user-facing change?

SDK readiness no longer blocks on replicated.app. Integration mode caches the license locally and falls back to cache when upstream is unreachable. Opt out with `requireUpstreamOnStartup: true`.

[bennyyang11]

bugbot run

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 53c18b9. Configure here.}