Background processes from old app versions break updates — critical #3605

Problem

When users update OpenHuman, stale background processes from the previous version remain running. These interfere with the new version, causing login failures, connection issues, and general breakage. Users have to manually kill processes or restart the machine. This was tagged as Critical priority in internal docs.

Root cause

The installer/updater doesn't cleanly terminate all child processes before replacing the binary. Old PIDs conflict with new ones.

Suggested fix

The update flow should:

1. Kill all openhuman-related processes before installing
2. Or the new version should detect and handle stale PIDs gracefully
3. Or use a version-lock on background services so old processes refuse to run if a newer binary is present

Evidence

• Notion wiki summary documenting "Critical high-priority background process bug"
• "App error requiring restart — users need to restart or clear data — requires immediate patch"
• Multiple sentry reports of CoreRpcError "Failed to fetch" and connection failures