Metadata GDSC annotation improvements

[Lilly-May]

Description of changes

• Fixed the logger error message in the cell line metadata annotation. The code will still fail, but now the informative logging message will be shown.
• The parameter specifying the GDSC dataset can now also be provided as a string. Feel free to reject this change, as I might be the only user making the mistake of providing it as a string instead of an integer. The latter causes the method to always use GDSC dataset 2 without specifically stating it, which might be overlooked by the user. I also added a test that the parameter is set to 1 or 2 and otherwise throw an error.
• The most important change is in the annotate_from_gdsc method, which now uses the correct GDSC data to compute. Previously, the code would fail for me if GDSC2 was used.
• Adapted the Statsmodel example. I believe an old class name might have been used here. It's still not a working example as the adata is not defined, but I think it is much easier for users to understand now.

[Zethson]

@Lilly-May can you change this so that we use string literals for the dataset names? Maybe Literal["gdsc_1", "gdsc_2"] or something of that sort? Then we can merge it

[Lilly-May]

Sure! But you still want it to work with integers as well, right? So something like Literal["gdsc_1", "gdsc_2"] | Literal[1, 2]?

[Zethson]

I'd drop support for integers