Clean up posterior predictive sample methods

[martinkim0]

• Remove unused library_size argument (could add this back later once it is implemented) in VAE.sample
• Add typing and additional documentation in VAE.sample
• Clean up tensor shape annotations in VAE.sample and RNASeqMixin.posterior_predictive_sample
• Use Python 3.9+ annotations in scvi.model.base._rnamixin
• Clean up code in RNASeqMixin.posterior_predictive_sample

[codecov]

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (7523a30) 89.18% compared to head (4ca3f35) 89.18%.

Files	Patch %	Lines
scvi/model/base/_rnamixin.py	91.66%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2377      +/-   ##
==========================================
- Coverage   89.18%   89.18%   -0.01%     
==========================================
  Files         152      152              
  Lines       12487    12477      -10     
==========================================
- Hits        11137    11127      -10     
  Misses       1350     1350              

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

[martinkim0]

meeseeksdev backport to 1.1.x