z-score only using training data

[michaeldeistler]

We currently infer the mean and std of our data from all datapoints, including the validation set.

It is good practice to consider the mean and std as model parameters and therefore use only the training data to identify the mean and std.

See #393

[codecov-io]

Codecov Report

Merging #411 (d8030aa) into main (39d6954) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##             main     #411   +/-   ##
=======================================
  Coverage   82.77%   82.77%           
=======================================
  Files          50       50           
  Lines        3187     3187           
=======================================
  Hits         2638     2638           
  Misses        549      549           

Flag	Coverage Δ
unittests	82.77% <100.00%> (ø)

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

Impacted Files	Coverage Δ
sbi/inference/snle/snle_base.py	97.77% <100.00%> (ø)
sbi/inference/snpe/snpe_base.py	93.38% <100.00%> (ø)
sbi/inference/snre/snre_base.py	97.08% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 39d6954...d8030aa. Read the comment docs.

[janfb]

Looks good!