Re-add tests/ directory

[A5rocks]

Making this PR to see if this quick fix works.

I want to double check that source = ["."] doesn't work before merging this.

[A5rocks]

Hm that didn't work. 🤔

EDIT: nevermind it did work, I don't know why the coverage report step of the Cython tests didn't show me what I was looking for. I guess maybe I somehow skipped the line while reading...

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00000%. Comparing base (f8a51b6) to head (b871153).

Additional details and impacted files

@@               Coverage Diff               @@
##                 main        #3314   +/-   ##
===============================================
  Coverage   100.00000%   100.00000%           
===============================================
  Files             125          127    +2     
  Lines           19253        19261    +8     
  Branches         1304         1303    -1     
===============================================
+ Hits            19253        19261    +8     

Files with missing lines	Coverage Δ
src/trio/_core/_tests/test_thread_cache.py	100.00000% <ø> (ø)
src/trio/_core/_thread_cache.py	100.00000% <ø> (ø)

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[webknjaz]

The problem with source is that it's ambiguously selecting either importables or directories. And you don't really know what you'll get upfront. But with the CWD, it's always a dir. And source_pkgs specifically recognizes importables.

This usually works best with Codecov, and pytest --cov (without a value set).

But since something wasn't right in trio, I've been postponing the original merge.

I'd recommend comparing the coverage. xml files produced before the PR and after. I suspect the base package dir attribute is different with your change.

[A5rocks]

Note for a later search: for some reason test_clear_thread_cache_after_fork is timing out.

The problem with source is that it's ambiguously selecting either importables or directories.

Yeah I tried to add a / just to ensure this is fine... We'll see. Hopefully codecov failure was just because of CI flakiness.