refactor: update batch size and mcmc defaults.

[janfb]

Update default arguments

Training batch size is set to training_batch_size=200.

MCMC default method is set to `method = "slice_np_vectorized", i.e., vectorized slice sampling. This makes it possible to run multiple chains in parallel fast and on a single CPU.

MCMC chains is set to num_chains=20 (was =1), because this is fast when vectorized.

fixes #910

[codecov]

Codecov Report

Attention: Patch coverage is 90.90909% with 1 line in your changes missing coverage. Please review.

Project coverage is 76.87%. Comparing base (b3254ed) to head (867c10a).
Report is 4 commits behind head on main.

Files	Patch %	Lines
sbi/utils/diagnostics_utils.py	88.88%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1221      +/-   ##
==========================================
- Coverage   85.44%   76.87%   -8.57%     
==========================================
  Files         101      101              
  Lines        7941     7945       +4     
==========================================
- Hits         6785     6108     -677     
- Misses       1156     1837     +681     

Flag	Coverage Δ
unittests	76.87% <90.90%> (-8.57%)	⬇️

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

Files	Coverage Δ
sbi/analysis/sensitivity_analysis.py	19.13% <ø> (ø)
sbi/diagnostics/sbc.py	95.71% <100.00%> (-0.07%)	⬇️
sbi/diagnostics/tarp.py	82.92% <ø> (-14.64%)	⬇️
sbi/inference/base.py	89.30% <ø> (ø)
sbi/inference/fmpe/fmpe_base.py	94.33% <ø> (ø)
sbi/inference/posteriors/mcmc_posterior.py	85.93% <ø> (-0.77%)	⬇️
sbi/inference/snle/mnle.py	85.36% <ø> (-7.32%)	⬇️
sbi/inference/snle/snle_base.py	93.87% <ø> (ø)
sbi/inference/snpe/snpe_a.py	63.91% <ø> (-27.40%)	⬇️
sbi/inference/snpe/snpe_base.py	91.61% <ø> (ø)
... and 8 more

... and 18 files with indirect coverage changes

[michaeldeistler]

LGTM, thanks!