Add normal distribution modelling of scVI for handling log-normalized data

[HelloWorldLTY]

Hi, I have implemented a normal-distribution-based scVI to handle data after log-normalization. Would you please review it? I have tested the clustering performance of it and it looks good. Thanks.

[codecov]

Codecov Report

Attention: Patch coverage is 25.00000% with 3 lines in your changes are missing coverage. Please review.

Project coverage is 88.86%. Comparing base (3087178) to head (94bf5a7).
Report is 2 commits behind head on main.

❗ Current head 94bf5a7 differs from pull request most recent head e35819d. Consider uploading reports for the commit e35819d to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2775      +/-   ##
==========================================
- Coverage   88.90%   88.86%   -0.05%     
==========================================
  Files         161      161              
  Lines       13339    13343       +4     
==========================================
- Hits        11859    11857       -2     
- Misses       1480     1486       +6     

Files	Coverage Δ
src/scvi/model/_scvi.py	95.71% <ø> (ø)
src/scvi/module/_vae.py	94.44% <25.00%> (-1.13%)	⬇️

... and 1 file with indirect coverage changes

[martinkim0]

Hi, thanks for your contribution. Looks good so far - could you add the following please:

• A release note in CHANGELOG.md
• Tests
• Possibly a small benchmark notebook showing how it performs vs. other distributions

[HelloWorldLTY]

Hi, I have modified the changelog.

For the tests, I am a bit confused. It seems that I had I/O error, how to address it?

For the testing, shall I directly use the datasets from the tutorial? Do you have any recommendation? Thanks a lot.

[martinkim0]

Hi @HelloWorldLTY, thanks so much for this initial PR! I've merged #2780 that adds support for this (I updated the docstrings and changelog).

[HelloWorldLTY]

Thanks a lot.