Add Datasets read and write methods

[QRemy]

Change names of datasets serialisation methods
datasets.to_yaml() -> datasets.write()
datasets.from_yaml() -> datasets.read()

[cdeil]

@QRemy - Please when opening a PR, always set a project, milestone and assignee. Assignee is the guy you want to review and merge your PR, in this case @adonath or me is fine.

If you don't have permission to set those things
let me know and I'll add you to the Gammapy maintainers team.

[cdeil]

@QRemy - Thanks.

Please adapt this caller:

https://dev.azure.com/gammapy/gammapy/_build/results?buildId=2699&view=logs&jobId=d7d66435-5ae4-561a-5876-2cd4e19b3ce2&taskId=922dfe32-5b90-53bc-18db-0fa0ecbbd97b&lineStart=142&lineEnd=143&colStart=1&colEnd=1

[cdeil]

Probably for consistency, we should call make_path in read, just like you do in write in https://github.com/gammapy/gammapy/pull/2631/files#diff-8b37a3d39becd8c4acbc3951517b883fR182 and like we do in all our read methods.

What this will achieve is that variable names are expanded, i.e. we can use Datasets.read("$GAMMAPY_DATA/some/file.yaml") in tutorials and the env var will be expanded.
I'm not sure it's a good pattern, but we have it in basically all our read methods, so Datasets.read should work the same way.

[cdeil]

@QRemy - Thank you!

[cdeil]

Changed title to "Add Datasets read and write methods" which I think will be more useful for users to see in the changelog.

I'm merging this now. Obviously we don't use this enough in the docs. @AtreyeeS - Please try to use it in the first and second analysis tutorial and your benchmark scripts, and report if you find any issues.