handle fill_value datatype for string-array

[zqfang]

this PR is related to the issue here: kaizhang/anndata-rs#15

@LDeakin Can you help review this PR if you have time ? I can make further changes if it looks the right workaround

[LDeakin]

What zarr-python seemingly does with a 0 fill value is map it to an empty string "", but this maps 0 to "0". All you really need to do is check if the fill value is 0 (like in the bool/int workaround) and map to an empty string (like in the string/null workaround).

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.13%. Comparing base (d5a5021) to head (69e2270).
Report is 3 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #140      +/-   ##
==========================================
+ Coverage   82.10%   82.13%   +0.02%     
==========================================
  Files         173      173              
  Lines       24495    24499       +4     
==========================================
+ Hits        20112    20122      +10     
+ Misses       4383     4377       -6     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[LDeakin]

Thanks for the contribution! I minimised the change in order to not be overly permissive and added some tests.

In writing the tests, I noted a bug in zarr-python itself: zarr-developers/zarr-python#2792

[zqfang]

ahh, your implementation is much cleaner and better!