Consider tests migration to bazel test

[nettle]

This is Bazel solution, so we should try to utilize "dogfooding" and use bazel approaches as much as we can, and do not transform testing to python.

Scope

1. Study bazel test based solutions and approaches, e.g.:
   • https://bazel.build/rules/testing
   • py_test, sh_test
   • https://github.com/bazelbuild/rules_testing
   • How to test that Bazel my_test rule fails when it should?
   • Bazel rules to test Bazel rules
2. Consider bazel test based approaches for unit testing and FOSS testing
   • Also consider removing python-based approaches
3. Select bazel test based approaches
4. Make decision
5. Develop a plan

Definition of Done

1. Bazel test based approach is selected
2. Decision is made
3. Migration plan is developed

[furtib]

We have made progress on this.
In the current state, we still use pytest for:

• FOSS tests -- extreme performance hit; running all at the same time causes a timeout. test: Prototype Bazel-native FOSS tests #209 is an exception; let's move forward with that. ✅
• legacy tests -- no plans to convert
• external_repository tests -- test: Migrate external_repo tests to bazel #298
• compile_flags tests -- I want to remove these; this behaviour can be reproduced with skylib tests. test: integrate compile_flags tests into skylib test #318
• plist_res tests -- test: Migrate plist_res test into a bazel test #312 ✅
• cache tests -- refactor: Migrate cache tests to bazel test #313 ✅
• parse tests -- Todo

Running Bazel inside Bazel introduces a very heavy performance hit. We should avoid it, or only use it in very trivial cases.