[ENH] parallelization backends for grid and random search

[SimonBlanke]

From Original PR #150:

This PR introduces parallelization backends for GridSearch and RandomSearch.

Both estimators now allow to specify backend and backend_params in the constructor, which allows selection of a parallelization backend and configuration parameters for it - the default being "None", i.e., plain loop.

This uses the parallel utilities also used in sktime and skpro, with a mid-term plan to move these to scikit-base.

[fkiraly]

Great, thanks for the examples.

May I request that we merge #150 first and then #162? To have the correct by-line blame in the repository (e.g, for questions later)

[fkiraly]

actually, I noticed test times are going up substantially, and one job timed out 30min.

We should find out what is causing that. It does not seem to be the case in #150.

[SimonBlanke]

Great, thanks for the examples.

May I request that we merge #150 first and then #162? To have the correct by-line blame in the repository (e.g, for questions later)

okay, but the conflicts need to be resolved first.

[SimonBlanke]

actually, I noticed test times are going up substantially, and one job timed out 30min.

We should find out what is causing that. It does not seem to be the case in #150.

The only differences between these two branches are some examples. I suspect, that the CI tests from a different branch were running while this one started tests. This does not mean, that we shouldn't make the test pipeline faster. But this does mean, that this PR should not tackle this issue.

[fkiraly]

looks like it´s coming from ray - should we skip those tests for now?

[SimonBlanke]

looks like it´s coming from ray - should we skip those tests for now?

Yes, let's skip it for now. Could you open an issue on how to test ray without blowing up our test pipeline, so that we remember this?