Logit transform

[anastasiakrouglova]

I changed the way we are z_scoring the data. Now, instead of none, independent and structured, we have can also logit transform our input, to train in an unbound space:

    - `independent`: Applies standard z-scoring with independent dimensions.
    - `structured`: Applies standard z-scoring with structured dimensions.
    - `logit`: Applies logit transformation, the user has to specify the distribution (prior for NPE and some distribution over data if this wants to be used for NLE)
    - `none`: No transformation

This implementation also allows logit-transforming the samples for numerically stable training when many samples are outside the prior bounds.

[codecov[bot]]

Codecov Report

:x: Patch coverage is 96.36364% with 2 lines in your changes missing coverage. Please review. :white_check_mark: Project coverage is 82.54%. Comparing base (65ae720) to head (1ee797d). :warning: Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
sbi/neural_nets/net_builders/flow.py	97.72%	1 Missing :warning:
sbi/utils/sbiutils.py	90.90%	1 Missing :warning:

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1485      +/-   ##
==========================================
- Coverage   86.43%   82.54%   -3.90%     
==========================================
  Files         135      135              
  Lines       10927    10931       +4     
==========================================
- Hits         9445     9023     -422     
- Misses       1482     1908     +426     

Flag	Coverage Δ
unittests	82.54% <96.36%> (-3.90%)	:arrow_down:

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
sbi/neural_nets/net_builders/flow.py	97.57% <97.72%> (+6.57%)	:arrow_up:
sbi/utils/sbiutils.py	79.15% <90.90%> (-7.71%)	:arrow_down:

... and 26 files with indirect coverage changes

[gmoss13]

Hi @anastasiakrouglova! Is there an update from your end on the status of this PR, or any input from our end that could help at this stage?

[anastasiakrouglova]

Hi @janfb and @gmoss13, thanks a lot for your comments! I adjusted the code to your feedback. Let me know if I have to adjust anything else 🚀

[janfb]

@anastasiakrouglova thanks for the updates! 🙏

is this ready for another round of reviews?