Design work for supporting resume API for drain mode

[soninaren]

When performing a scale-in operation on a function app, Antares infrastructure will put the function app in drain mode. This will effectively make it so that the function app would stop fetching new events. However this leads to another issue if we the function app needs to scale out again very quickly. At that point platform could chose a worker that is new, or a worker that is potentially still asleep causing the event queue to build up and leads to the function app assigned to workers that are forever asleep. This risk is particularly elevated for stamps that have limited capacity.

To address this issue we need an API that starts the worker that was previously put in drain mode, allowing it fetch new events.

[fabiocav]

Moving this to sprint 111.

/cc @liliankasem

[liliankasem]

Closing this out as initial design doc is in and we created issues to track the next steps:

• [Resume API] Investigate restart API feasbility #7781
• [Resume API] Implement resume API endpoint in the functions host #7774
• Improve cancellation token experience for drain and shut down #7785