Self-Update: maintenance page does not auto-reload after restart

[Wiesenwischer]

Problem

After triggering a self-update, the maintenance page shows "Restarting with new version..." but does not automatically reload when the new container is ready. The user has to manually press F5 to continue.

This has happened multiple times.

Expected

The maintenance page should automatically detect when the new container is running and redirect to the dashboard.

Current Behavior

The page shows the spinner and version transition (e.g. v0.50.1 → v0.51) indefinitely. Manual F5 reload shows the updated version correctly.

Likely Cause

The auto-reload polling in the helper container's maintenance page either:

• Loses the connection when the old container is removed
• Polls the wrong endpoint
• Has a timing issue where the new container is ready but the page doesn't detect it

[Wiesenwischer]

Additional observation

After waiting ~2 minutes, the maintenance page transitions to "Update failed — Update is taking longer than expected". Clicking "Back to Dashboard" or pressing F5 shows the successfully updated v0.51 instance.

The update itself succeeds — the issue is purely in the polling/detection logic of the maintenance page served by the rsgo-updater helper container.

[Wiesenwischer]

Second issue: SemVer mismatch on retry

After the initial "Update failed" timeout, clicking "Retry update" shows:

"Version 0.51 is not newer than the current version 0.51.0"

Root cause: GitHub Release tag is v0.51 (2-part) but the Docker image version is normalized to 0.51.0 (3-part SemVer). The version comparison treats 0.51 as not newer than 0.51.0.

Fix options:

1. Always create 3-part tags (v0.51.0 instead of v0.51) — requires milestone naming convention change
2. Normalize version comparison to handle 2-part versions (0.51 == 0.51.0)
3. Version endpoint returns the tag as-is, not the normalized build version