Streamline Testing with pytest-cov and pytest Defaults

[WaelKarkoub]

Why are these changes needed?

This PR aims to simplify the testing process and reduce redundancy in the GitHub workflows.

This PR does the following:

• Migrates to pytest-cov: Instead of relying on the coverage package, this PR adds pytest-cov plugin. pytest-cov works well with pytest-xdist, allowing us to run tests concurrently if the CI environment supports it.
• Unifies test execution: The workflows have been updated to use the pytest command directly, replacing the previous coverage -a -m invocation. This unifies the testing process and simplifies the workflows.

Update 1:

Copy pasting from discord:

There's seems to be a silent failure in master:

• with coverage the github action step passes even though the tests are failing.
• with pytest-cov, the github action step fails correctly

The environment variable AUTOGEN_USE_DOCKER was not being used properly in these tests, so I added some fixes:

• added a fix in test_commandline_code_executor.py,
• added a fix in test/test_code_utils.py

Update 2:

• There is a bug in Disc cache that doesn't handle windows path correctly, so I added a fix there as well.
• This also led to a CI a bug PermissionError: [WinError 32] The process cannot access the file because it is being used by another process: 'C:\\Users\\RUNNER~1\\AppData\\Local\\Temp\\tmpcq4z30pm\\42\\cache.db', for windows-2019 and python 3.12 for image gen cache test. I added a skip for that condition till I figure out the env.

Update 3:

• Dropped openai version to openai>=1.3
• Mypy ignores openai_utils.

Related issue number

Checks

• I've included any doc changes needed for https://microsoft.github.io/autogen/. See https://microsoft.github.io/autogen/docs/Contribute#documentation to build and test documentation locally.
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.

[WaelKarkoub]

@sonichi we're almost there! test/oai/test_client_stream.py::test_chat_tools_stream is a flaky test (even for me), and it doesn't test what it's supposed to be testing. I added a fix for it as well

[sonichi]

@WaelKarkoub Thanks. There is some notebook test failure which I will fix.
@ekzhu @thinkall this is one reason we don't want to use a single config in the notebook.