fix: rejection posterior should use theta_transform

[janfb]

bug: RejectionPosterior was not using a theta_transform when sampling with a constrained prior. This is fixed in this PR by removing the optional kwarg in the RejectionPosteriorParameters config class and instead passing it on during build_posterior by default and also using it when sampling with reject_outside_prior=True.

related to #1718

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 87.99%. Comparing base (49fa029) to head (6e8adbb).
⚠️ Report is 2 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1827      +/-   ##
==========================================
+ Coverage   87.95%   87.99%   +0.03%     
==========================================
  Files         140      140              
  Lines       12795    12985     +190     
==========================================
+ Hits        11254    11426     +172     
- Misses       1541     1559      +18     

Flag	Coverage Δ
fast	82.77% <ø> (?)

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

Files with missing lines	Coverage Δ
sbi/inference/posteriors/posterior_parameters.py	83.73% <ø> (+0.40%)	⬆️
sbi/inference/posteriors/rejection_posterior.py	78.18% <ø> (ø)
sbi/inference/trainers/base.py	94.10% <ø> (ø)

... and 1 file with indirect coverage changes

[manuelgloeckler]

Thanks for fixing!

Not sure why, but one test on signatures fail posterior_parameters_test which checks for theta_transform but this got removed herein so the test should be adapted before merging.

Edit: I fixed it.