Rudaux autoext does weird things when reg dates are updated erroneously

[trevorcampbell]

Dec 21 of the Fall 2021 semester, after final exam -- For some reason some of my students are appearing with registration dates of Dec 20. Rudaux then tries to give them an extension to Dec 27.

We should include a config parameter that specifies the final course add date so that even if a student's reg date is erroneously updated late in the course, rudaux won't give them a weird extension.

For example, in the autoext module:

[2021-12-21 16:01:41+0000] INFO - dsci100-001-autoext - prefect.get_latereg_overrides[0] | Student Firstname Lastname needs an extension on assignment worksheet_01
[2021-12-21 16:01:41+0000] INFO - dsci100-001-autoext - prefect.get_latereg_overrides[0] | Student registration date: 2021-12-20T23:18:07+00:00    Status: active
[2021-12-21 16:01:41+0000] INFO - dsci100-001-autoext - prefect.get_latereg_overrides[0] | Assignment unlock: 2021-09-09T22:30:00+00:00    Assignment deadline: 2021-09-12T06:59:59+00:00
[2021-12-21 16:01:41+0000] INFO - dsci100-001-autoext - prefect.get_latereg_overrides[0] | Current student-specific due date: Sat 2021-09-11 23:59:59 from override: False
[2021-12-21 16:01:41+0000] INFO - dsci100-001-autoext - prefect.get_latereg_overrides[0] | Late registration extension date: Mon 2021-12-27 23:59:59
[2021-12-21 16:01:41+0000] INFO - dsci100-001-autoext - prefect.get_latereg_overrides[0] | Creating automatic late registration extension.

I haven't fixed the autoext module, but I have halted it (since I don't need it any more anyway for this semester).

Suggested fixes

• get central not to screw with registration dates (....not going to happen)
• add a config param that gives rudaux a date after which it ignores registration
• have rudaux itself keep an internal reg date for students based on the first time it sees them in the student list (I like the idea but it is against the design philosophy of not keeping a state...but then again for the new version of rudaux there will be tonnes of implicit state, so honestly this might be best....)