set meta by pd.DataFrame

[danielhuppmann]

Please confirm that this PR has done the following:

• Tests Added
• Documentation Added
• Description in RELEASE_NOTES.md Added

Description of PR

In the existing implementation of set_meta(), passing over a non-valid index (e.g., a pd.DataFrame) was simply ignored without warning or error. This PR is a clean-up in the sense that if there is an index arg, it will be used and an error raised if it cannot be coerced in a meaningful way.

In addition, the function now also takes a pd.DataFrame with columns ['model', 'scenario'] as valid index arg.

[gidden]

Ok, I think I see the point here, but one question. The test as is has the following lines

    df = pd.DataFrame([
        ['a_model', 'a_scenario', 'a_region1', 1],
    ], columns=['model', 'scenario', 'region', 'col'])
     meta_df.set_meta(meta=0.3, name='meta_values', index=df)

Would the existing implementation work if the user provided the following?

    df = pd.DataFrame([
        ['a_model', 'a_scenario', 'a_region1', 1],
    ], columns=['model', 'scenario', 'region', 'col'])
     meta_df.set_meta(meta=0.3, name='meta_values', index=df[['model', 'scenario']])

[gidden]

do we need to check if index is None here?

[gidden]

Ok, answered my own question by looking back through. Looks great to me!