tests: switch from GTest to Catch2

[henryiii]

Finally dumping GTest. There are lots of reasons, such as

• Better compiler support. GTest has killed off GCC 4.8 (CentOS 7, manylinux1). I already had to drop GCC 4.7 support due to GTest, not doing it again. (Not that anyone cared about 4.7. 4.8 though, is a major one). Clang <5 also was dropped by GTest. Also GTest doesn’t seem to handle C++20.
• Better support. Catch2 (and probably doctest) respond to user issues faster, GTest is a bit internal to Google. (Better than it used to be)
• Real versions, and changelog. Live at HEAD is fine, but should not be forced, and changelogs let you track what happens from a high level.
• Simpler tests. Boilerplate things (templates, repeated code, generation) are simpler, and look prettier too.
• Matchers built-in, instead of requiring GoogleMock, which is an extra pain (this is a key reason as to why why I'm going with Catch2 over doctest).
• More powerful runner if we do want to use it.
• Nicer output.

Caught a few minor issues in the tests, like a duplicated type in the type parametrization. Much of the conversion done by https://gist.github.com/henryiii/7dc93c6bd8404b346daa2647e72fbd55, based on https://anteru.net/blog/2017/from-google-test-to-catch/.

Fixing the CMake config check by using an action + composite action instead of docker.

[codecov]

Codecov Report

Merging #574 (d44cf04) into master (34c4310) will not change coverage.
The diff coverage is n/a.

❗ Current head d44cf04 differs from pull request most recent head 73ce349. Consider uploading reports for the commit 73ce349 to get more accurate results

@@            Coverage Diff            @@
##            master      #574   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           12        12           
  Lines         3780      3780           
=========================================
  Hits          3780      3780           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 34c4310...73ce349. Read the comment docs.