Update Store with Version Info

[marcw-replicated]

What does this PR do?

Problem: In HA deployments (replicaCount > 1), SDK pods initialize version-related fields (channel sequence, version label, etc.) from the config file once at startup and never update them. If a pod survives a rollback or restarts at a different time than its peers, it can end up with a stale channel sequence in memory. Both pods then pass the canReport check and send heartbeats every 4 hours at slightly different times, causing the vendor backend to see the instance oscillating between two versions and firing repeated instance.upgraded notifications.

Fix: Added a Kubernetes informer (pkg/secretwatcher) that watches the replicated Secret for updates. When the Secret changes (e.g., after a Helm upgrade), all version-related fields (channelID, channelName, channelSequence, versionLabel, releaseSequence, releaseCreatedAt, releaseNotes) are updated in the in-memory store immediately. This ensures all running replicas converge on the correct version data without requiring a pod restart.

Changes:

• Added Set* methods to the Store interface and InMemoryStore for all version-related fields
• Added pkg/secretwatcher package implementing a Secret informer using ListWithContextFunc/WatchFuncWithContext
• Wired the secret watcher into bootstrap.go, starting it before the heartbeat

Does this PR introduce a user-facing change?

NONE

[laverya]

This isn't needed due to the existing "include config digest in annotations, and report only if latest pod generation" pattern