Revisit some issues with scheduled runs and upstream tests collection

[jaimergp]

Description

Fix a couple issues where a failed scheduled run wouldn't trigger a new issue notifying us of the problem. This PR also revisits the pytest collection from upstream conda/conda so we have more control over it; it uses a pytest plugin (dev/collect_upstream_conda_tests), which needs to be installed in the same environment as pytest.

Checklist - did you ...

• Add a file to the news directory (using the template) for the next release's release notes?
• Add / update necessary tests?
• Add / update outdated documentation?

[jaimergp]

This PR is also uncovering the errors we see on main, which started happening on June 16th for conda-forge. Note their equivalents with defaults do pass.

• Linux, macOS and Windows:
  • conda/tests/core/test_solve.py::test_force_remove_1
  • conda/tests/core/test_solve.py::test_aggressive_update_packages
  • conda-libmamba-solver/tests/test_modified_upstream.py::test_pinned_1
• Linux only:
  • conda/tests/test_create.py::test_conda_pip_interop_dependency_satisfied_by_pip
  • conda/tests/test_create.py::test_conda_pip_interop_pip_clobbers_conda
  • conda/tests/test_create.py::test_install_tarball_from_local_channel

[jaimergp]

Turns out the failures were #186 hitting again after 1.4.4 was released on June 16th. I've added a pytest plugin (WIP) to collect tests conditionally depending on what's available. This is also cleaner than overriding the pyproject.toml / setup.cfg config in conda/conda.

[jaimergp]

pre-commit.ci autofix

[costrouc]

All green 😄. @jaimergp does this deserve news?

[jaimergp]

@costrouc, yay! In that case can you do a quick review? Thanks!

[jaimergp]

And it's an internal thingy for our own test suite, not sure it needs a news entry. I edited the PR description just in case.