[The Framework] Background jobs

[brillout]

Implements:

• Listeners
  • Error in production (Sentry, Cloudfalre, ...) => triggers agent
  • CI is red on main (GitHub) => triggers agent
• The usage max-out idea:
  • Check the current usage limit + when the limit resets
    • If usage limit reached (with a configurable margin) => abort, don't do any maintenance
    • Make sure to also check the usage limit of the current default model => use a fallback model if usage limit is reached (with a configurable margin)
  • If there's capacity, then check the latest commits of all The Framework repos that weren't reviewed by the maintenance loop yet => apply the maintenance loop
• Autopilot mode (fully automatic development cycle, including feature requests)
• More?