[python-package] [docs] Expand class docs for Dataset

[Plenitude-ai]

Fixes #6511

As discussed with James Lamb in #6511 (comment), a bit more documentation about the way the lgb.Dataset is actually constructed, with a particular representation of data, could be interesting for users.

[jameslamb]

Thanks very much for this!

As I mentioned in #6511, I'll push a commit here with the corresponding change for the R package, and then we can ask for other maintainers to review.

---

In the future when you use the pull-request-from-a-fork contributing process, I strongly recommend creating your own branch instead of using main / master. See #6501 (review) for some reasons and https://github.com/scipy-conference/scipy2024_sprints_first_PR/blob/main/prs_and_reviewing.md for some advice on that.

[jameslamb]

I've also put "fixes #6511" in the description, so that will be automatically closed when we merge this. I believe this answers your final question there.

dismissing my review because I pushed commits here

[jameslamb]

Thanks again @Plenitude-ai . I've pushed commits updating the formatting and adding the corresponding R changes. Please follow the instructions in #6558 (comment) (sorry, it's a Microsoft requirement), and one of the other maintainers here will review when they have time.

[Plenitude-ai]

Thanks again @jameslamb for your guidance. I realize it's a (very) modest one, but this is my first contribution to an open-source project so it's a great pleasure !
Thanks for your different advices on good practices in this process, i'll definitely follow them in the future.
@microsoft-github-policy-service agree company="Qwant"

[jameslamb]

@Plenitude-ai wonderful to hear it, thanks for choosing LightGBM for your first contribution! We really appreciate the help, and hope you'll consider contributing more in the future.

[StrikerRUS]

@Plenitude-ai Thanks a lot for your contribution!
And thanks @jameslamb for helping with R-package!