Numpy now uses a version cap for setuptools in their `pyproject.toml`,
which defies a bit the value of including it in the integration tests.

As revealed in a conversation with `numpy` maintainer, it seems that
they not plan to remove this cap (maybe only update it from time to
time, only if necessary). Moreover they are also studying other
build backends.

So I think that now the best thing to do would be stop trying to build
numpy sdists with the current version of setuptools, otherwise we risk
to break this test.

Maintain `requires` order in METADATA.

Some tests are running the build process using setuptools own directory
as cwd. This impacts the build process and also leave behind artifacts
produced during tests (like .egg-info folders)

Instead of re-building/installing setuptools from the source tree
every time, the tests now rely on the venv, wheel and sdist fixtures
(the venv fixture is populated from sdist/wheel).

Moreover migrate `test_virtualenv` to use `jaraco.envs`
(so it uses the same libraries ad `test_distutils_adoption`).

… and change it to install the pre-build setuptools wheel (fixture)
instead of installing from the source tree

They should be build once for each session and be able to be re-used in
parallel (assuming read-only) for all tests.

This is useful when dealing with virtual environments