Update _anndata.py - Clarify a bug & Add Subsetting Functionality

[jwalewski]

Changes

The one change I wish to make relates to the error handling present in an edge case to do with merging:

• Currently, if one merges and there are no matching observations and there is no id_length given to clean the observation names, a traceback is triggered that says id_length is referenced before assignment. Since this isn't very informative of the real issue (that no id_length value was given and that there were no matching indices) I am proposing a minor change where the error is raised with a more descriptive message.

Bug fixes

• None other than the change.

New

Throughout my time using scvelo I really wished that it was possible to directly subset based on observations being equal to a specific value (as compared to taking entire variables), so I wrote a subset_by_value function for my own work. I think it is helpful enough to be included in the repository as a whole.

Let me know if there is any other functionality that should be included in it and/or what tests should be run before it is merged into the main branch.

Linked issue

#1183