Pass kwargs through task to get_dataset

[knyazer]

Allows to follow the directions in the FutureWarning Starting from Version 0.15 `download_data`, `download_qualities`, and `download_features_meta_data` will all be ``False`` instead of ``True`` by default to enable lazy loading.

Reference Issue

None

What does this PR implement/fix? Explain your changes.

When doing something like

dataset = task.get_dataset()

which takes no parameters one gets the warning. Now it takes any parameters that the datasets.get_dataset takes, thus allowing to set the optional parameters.

How should this PR be tested?

Test that the suite passes, I guess; there is not really a short way to test.

Any other comments?

Maybe kwargs-passing behaviour should be added to all the other methods in the OpenMLTask?

(this PR might be noise again, though I checked that the issue is present on dev, sorry if I am wrong though)

[PGijsbers]

Thanks for the contribution! I think it's fair to allow this, but it needs a small improvement for the documentation. I don't think it is necessary to add tests for this functionality.

[knyazer]

Btw, are you in need of any maintainers/collaborators? I'm happy to be spending quite a bit of time supporting openml, since I work on automl, and frequently use the library, and I have a bit of free time this year (part-time job, so I can work on random projects quite a bit).

And yeah, it seems like there are a lot of low-ish priority feature requests in the issues, and you are effectively the only maintainer..

So yeah, if you are happy with that I can just start implementing/fixing random stuff pretty actively 😉

[PGijsbers]

Absolutely! It would be best if you could join our Slack server (see https://openml.org/about) for general contact and maybe discussing the kind of issues we currently prioritize. For discussing specific issues/prs, GitHub is the main platform. We also do monthly engineering calls that are open. Every first monday of the month at 17.00 CEST (when daylight saving is applicable in Europe, otherwise CET).
We always appreciate the help!

[knyazer]

Alright, disallowing positional arguments while there are zero of them is useless; I just put slash there to be more explicit, as a sort of self-documentation :)

[knyazer]

And yeah, thanks for fixing it, the back and forth through the GitHub is a lot slower than changing two chars in the commit