Wrap 1D PyTorch distributions

[michaeldeistler]

Closes #1285 an #1284

1D pytorch distributions such as torch.distributions.Exponential, .Uniform, or .Normal do not, by default return any sample or batch dimension. E.g.:

dist = torch.distributions.Exponential(torch.tensor(3.0))
dist.sample((10,)).shape  # (10,)

sbi will raise an error that the sample dimension is missing. A simple solution is to add a batch dimension to dist as follows:

dist = torch.distributions.Exponential(torch.tensor([3.0]))
dist.sample((10,)).shape  # (10, 1)

Unfortunately, this dist will return the batch dimension also for `.log_prob():

dist = torch.distributions.Exponential(torch.tensor([3.0]))
samples = dist.sample((10,))
dist.log_prob(samples).shape  # (10, 1)

This will lead to unexpected errors in sbi. The point of this PR is to wrap those batched 1D distributions to get rid of their batch dimension in .log_prob().

[codecov]

Codecov Report

Attention: Patch coverage is 86.36364% with 3 lines in your changes missing coverage. Please review.

Project coverage is 78.36%. Comparing base (299854e) to head (34b081e).
Report is 8 commits behind head on main.

Files with missing lines	Patch %	Lines
sbi/utils/user_input_checks_utils.py	85.00%	3 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #1286       +/-   ##
===========================================
- Coverage   89.24%   78.36%   -10.89%     
===========================================
  Files         119      119               
  Lines        8695     8727       +32     
===========================================
- Hits         7760     6839      -921     
- Misses        935     1888      +953     

Flag	Coverage Δ
unittests	78.36% <86.36%> (-10.89%)	⬇️

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

Files with missing lines	Coverage Δ
sbi/utils/user_input_checks.py	76.56% <100.00%> (+0.24%)	⬆️
sbi/utils/user_input_checks_utils.py	89.17% <85.00%> (-0.61%)	⬇️

... and 33 files with indirect coverage changes

[gmoss13]

Thanks @michaeldeistler! I left a couple of minor comments, please have a look but I am happy to merge this when the pyright issue is fixed. By the way, I think you meant that this PR addresses #1283 instead of #1284 :)