Add scheduler for scan tasks

[marcelldls]

Why:

• Set intervals rather than delays (avoid drift, add precision) currently the property name period is a bit disingenuous
• Behavior for scan tasks that take longer than their interval. Now sets to skip - Get a useful warning message: Execution of job "readback_position (trigger: interval[0:00:00.010000], next run at: 2024-10-29 15:13:39 GMT)" skipped: maximum number of running instances reached (1)

Possibly useful options not used:

• Dynamically add/remove tasks (Destroy tasks on exit?)
• Pause tasks (during reconnect?)
• Custom triggers (e.g. mock trigger so that we can step in testing, Cron type triggers)
• Grace period

Uncertain:

• Handle the log messages produced

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.21%. Comparing base (bdab4fa) to head (edd8d49).

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #68      +/-   ##
==========================================
- Coverage   82.75%   82.21%   -0.55%     
==========================================
  Files          23       23              
  Lines         928      911      -17     
==========================================
- Hits          768      749      -19     
- Misses        160      162       +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[GDYendell]

This seems good! I agree with both of your "Why"s.

I am not familiar with apscheduler. Is that considered the standard solution for this? Are there alternatives we should look at?

[marcelldls]

Reasons for:

• Support for asyncio
• Actively maintained, 6300 stars, 300000 daily downloads (Seems to be usually used as a plugin in web applications django-apscheduler, Flask-APScheduler, etc.)
• It seems to be popular enough that there are plenty of tutorials available https://www.youtube.com/watch?v=sRnsR-T0Lxc

Possible doubts:

• Maybe too heavy a library for what we need?
• Overhead on polling performance?

An alternative could be:

• DIY
  • Keep track of active periodic tasks and do our own logic and handling, use a call back to pop task when done
  • Use a while <attribute> to stop execution
  • Move sleep into gather()? like in https://stackoverflow.com/questions/37512182/how-can-i-periodically-execute-a-function-with-asyncio
• aiojobs? 850 stars, 5000 daily downloads. Pretty minimal so possibly a middle ground