DEV/MNT: migrate to latest standards with pyproject.toml [LCLSPC-603]

[klauer]

Description

• Upgrading by way of https://github.com/pcdshub/pcds-python-migration-tools
• pcdsdaq has been migrated to use setuptools-scm, replacing versioneer, as its version-string management tool of choice.
• pcdsdaq has been migrated to use the modern pyproject.toml, replacing setup.py and related files.
• Older language features and syntax found in the repository have been updated to Python 3.9+ standards by way of pyupgrade.
• Sphinx 6.0 is now supported for documentation building. docs-versions-menu replaces doctr-versions-menu and doctr usage for documentation deployment on GitHub Actions. The deployment key is now no longer required.
• pyupgrade was used to update annotations, though there might not have been many here
• pycln was used to clean up imports

Motivation and Context

• Migrating because we want to migrate our projects from Travis CI for one
• Second migration attempt after nudging #108

How Has This Been Tested?

• GitHub Actions

Where Has This Been Documented?

This PR and Jira LCLSPC-603

[klauer]

A few issues:

• psdaq-control-minimal isn't/won't be on pypi any time soon, so pip builds will fail
• There's a reliance on type hints where we run into a similar issue as in atef: int | float | None is fine with from __future__ import annotations but is problematic when we try to evaluate those type hints in Python < 3.10.
  • https://github.com/klauer/pcdsdaq/blob/47abeb2703211b2d7d10d84713a4410b2ad2cf71/pcdsdaq/daq/interface.py#L229-L268
• There's one failing test in 3.10/conda: https://github.com/pcdshub/pcdsdaq/actions/runs/4367132731/jobs/7638055269
• 3.11/conda lightpath strikes again

[klauer]

I reverted some type hints to fix the related issue for Python 3.9.

I'm not sure if the currently-failing test should be failing since Travis CI just didn't run on this repo previously. I'll ask for Zach to review and see what he thinks.

[ZLLentz]

Based on my docstrings and comments, the failing test is an actual failure (barring test setup error possibilities, etc.) I can make an issue to revisit:

• The failing test
• Make psdaq an optional dependency and skip tests that need it when missing

[klauer]

OK, bypassing branch protection to merge with the acknowledgement that there is follow-up work to be done (#117, #118)