Tweak sample counts and other details in tests

[mhauru]

This PR makes a number of small changes to the tests

• Lowers the sample counts of many tests.
• Adds a few comments, including some TODO notes saying we should add an explanatory comment.
• Uses StableRNGs more. Still not using them everywhere, but I added them in many places where stability matters.
• Removes unnecessary calls to Random.seed!. Every @testset resets the seed anyway.
• Removes some dead or commented out tests that weren't doing anything.
• Adds verbose = true to many of the top-level @testsets, to get some timing numbers even when tests pass.
• Adds info prints in the beginning of some of the longer-running testsets, saying which test is starting.
• Adds some more subtestsets to some of the longer-running ones, to get more granular info about what fails.
• Fixes one commented-out test to actually do something.

This started with me thinking that since we do a lot of samples in many tests, and our test suite is quite slow, clearly we should reduce the sample counts if we can. What I learned in the process is that most of the tests that do sampling are negligble compared to the tests that use particle methods. Anything involving PG/CSMC tends to take 10-100 times longer than comparable tests with other samplers.

In the end I've tried to adjust iteration counts only when either the count truly doesn't matter (we are not checking stats of the chain), or when doing so actually has a noticable, though not necessarily huge, impact on run time. I may have slipped a few times, and reduced something that didn' really make a difference. Also, when reducing iteration counts I tried to check that they pass if I pick a few different seeds, to make sure I'm not making them very brittle to future changes.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 86.41%. Comparing base (c0a4ee9) to head (544f9be).
Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #2433   +/-   ##
=======================================
  Coverage   86.41%   86.41%           
=======================================
  Files          22       22           
  Lines        1575     1575           
=======================================
  Hits         1361     1361           
  Misses        214      214           

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

[coveralls]

Pull Request Test Coverage Report for Build 12321043702

Warning: This coverage report may be inaccurate.

This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.

• For more information on this, see Tracking coverage changes with pull request builds.
• To avoid this issue with future PRs, see these Recommended CI Configurations.
• For a quick fix, rebase this PR at GitHub. Your next report should be accurate.

Details

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-1.8%) to 76.801%

---

Totals
Change from base Build 12110228665:	-1.8%
Covered Lines:	1205
Relevant Lines:	1569

---

💛 - Coveralls

[mhauru]

I'm hoping CI will now pass, except for Mooncake stack overflows. See updated OP for what this PR does.

[penelopeysm]

Also, if you could bump Mooncake compat to 0.4.61, that would hopefully take care of CI 👍

[penelopeysm]

I can approve and merge once CI passes :)

[penelopeysm]

Some Ubuntu tests are looking like they will time out, but I'm going to merge on the basis that the tests pass on Windows and macOS and the root cause for the timeout is known (TuringLang/DynamicPPL.jl#750) and will presumably be fixed soon (TuringLang/DynamicPPL.jl#751).