TST: Attempted test suite fixups

[tangkong]

Description

Attempting to fix some intermittent test suite failures. The crux of the difficulty here is that the suite consistently passes locally, failing only intermittently on GHA CI

Motivation and Context

• Uses / vendors Typhos' WeakPartialMethodSlot for applying callbacks within ActionRowWidget

• In one case opted to try making the slots submitted to BusyCursorThread actual methods, rather than dynamically created functions

• When pcdsutils is updated, we can remove the vendoring here and rely on pcdsutils' versino

• There is still a very rare test suite failure, but I cannot reproduce it enough to address it. For the curious, it seems to fail at get_curr_value, the epics-get portion of ActionRowWidget

  • While this is the one place we chose to make the callbacks methods rather than partial slots, I think the issue is more likely due to late timeouts than widgets being garbage collected early

How Has This Been Tested?

GHA, things pass pretty consistently locally

Where Has This Been Documented?

This PR

Pre-merge checklist

• Code works interactively
• Code contains descriptive docstrings, including context and API
• New/changed functions and methods are covered in the test suite where possible
• Code has been checked for threading issues (no blocking tasks in GUI thread)
• Test suite passes locally
• Test suite passes on GitHub Actions
• Ran docs/pre-release-notes.sh and created a pre-release documentation page
• Pre-release docs include context, functional descriptions, and contributors as appropriate

[tangkong]

Since I mistakenly left an assert False statement, gha now sees what I was seeing locally. While all the tests fail, some of them fail + "attempt to access deleted object"

py3.11, pip fails when trying to teardown some PageWidgets, though test suite completes

py3.9, conda fails with a straight up segfault, suite does not complete

[tangkong]

A slightly better result, though failing the first time (py3.10, conda)

Edit: Through 4 retries not once did the entire matrix pass. I declare this "Not good enough"

[tangkong]

Replacing all partial callbacks with WeakPartialMethodSlot cleans up the test suite quite nicely.

I may try to add the class to pcdsutils before I merge this, for the sake of not having to return to this

[tangkong]

I'm not sure if I'm supposed to keep hard references to the WeakPartialMethodSlot object. I got a single segfault testing locally, but haven't been able to get one since, so I'm not sure if it's actually a problem. The GHA test suite seems to be passing without issue, so that's not really helpful

[tangkong]

This PR has become somewhat of a sandbox. I've tried storing references to the created WeakPartialMethodSlot instances, which seems to have helped. I've also (maybe naively) added the arguments received from the signal to the _call method, which doesn't seem to break anything.

[klauer]

I'm not sure if I'm supposed to keep hard references to the WeakPartialMethodSlot object

Looks like you figured it out - the answer is yes 👍

[tangkong]

This is ready to be reviewed and merged, with pcdsutils getting WeakPartialMethodSlot

[ZLLentz]

I think all these changes are clean and well-motivated
One nitpick about type annotations

[ZLLentz]

This bit is confusing from a type hinting perspective and probably gives some of the stricter type checking libraries pause. It's saying "if this thing that is always an EpicsSignalBase is None..."

Maybe this type should be EpicsSignalBase | None and maybe it should be hinted in the object init or in the class definition itself instead of in this method?

[klauer]

It might also just be a regular signal, right?

[tangkong]

Turns out atef.Procedure.Target.to_signal() -> Optional[EpicsSignal] as written currently, so I'll change it to match and move it to the init.

Edit: I suppose that should also just be a base signal as well.

[klauer]

Same as @ZLLentz - a bit of class body type hints are probably in order here.

[klauer]

It might also just be a regular signal, right?