Enables MG python tests using a single-GPU LocalCUDACluster in CI

[rlratzel]

closes #3413

This PR enables MG python tests using a single-GPU LocalCUDACluster in CI by setting the DASK_WORKER_DEVICES to 0. This does not affect SG tests, which continue to run in CI as they previously did.

PR #3540 added support for DASK_WORKER_DEVICES to the pytest fixture used in python MG tests, allowing CI scripts (and developers) to restrict workers to specific devices, which should now allow single-GPU CI runs to cover the MG/dask code paths in python.

Note: despite the changes here to now run python MG tests, it should be noted that this isn't actual multi-GPU test coverage in libcugraph since multiple GPUs are not communicating and the code to set up comms, distribute graph data, etc. is not being exercised using >1 GPU.

[VibhuJawa]

Approving but we should wait before merging to see amount of time added , CC: @rlratzel

[rlratzel]

I ran the MG tests using DASK_WORKER_DEVICES=0 locally and observed they took ~20 minutes, meaning our CI runs will be at least 20 minutes longer now assuming the CI system is equal to or slower than my system.

This isn't ideal but it's a decent tradeoff for getting the additional coverage.

A few notes:

• This PR does not remove any of the conditional skip markers that existed that check for >1 GPU, so there's still some missing coverage. But even with those in place, we're still adding significant coverage. Those will be removed in a future PR.
• We still have a lot of testing tech debt to clean up, so making the behavior of using the mg marker be "uses >1 GPU" and sg be "uses only 1 GPU" without setting DASK_WORKER_DEVICES will be done as part of closing Python tests should have a shared set of fixtures #1347

[BradReesWork]

/merge