fix sparse backed data loading

[adamgayoso]

Backed data currently doesn't work if X is sparse due to a minor bug.

I tested this and it works, with 1200 genes it's about twice as slow to load backed sparse. We can expose setting data loader num_workers to improve speed.

[codecov]

Codecov Report

Base: 90.45% // Head: 90.46% // Increases project coverage by +0.00% 🎉

Coverage data is based on head (c897fad) compared to base (c4c3c22).
Patch coverage: 100.00% of modified lines in pull request are covered.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1824   +/-   ##
=======================================
  Coverage   90.45%   90.46%           
=======================================
  Files         130      130           
  Lines       10337    10337           
=======================================
+ Hits         9350     9351    +1     
+ Misses        987      986    -1     

Impacted Files	Coverage Δ
scvi/__init__.py	87.50% <ø> (ø)
scvi/module/_scanvae.py	79.82% <ø> (ø)
scvi/module/_vaec.py	85.71% <ø> (ø)
scvi/utils/_track.py	72.72% <ø> (ø)
scvi/dataloaders/_anntorchdataset.py	92.64% <100.00%> (+1.47%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.