Adds possibility to specify batch_norm, bias and batch_norm_scale for each layer

[MaxFBurg]

To specify batch norm and its properties for each layer individually, this PR changes the attributes batch_norm bias batch_norm_scale and the according class members to lists.

Backwards-compatibility is ensured.

Also includes #215 which should be merged first.

[codecov]

Codecov Report

Attention: 19 lines in your changes are missing coverage. Please review.

Comparison is base (d66f372) 7.35% compared to head (3c0c412) 7.38%.

Files	Patch %	Lines
neuralpredictors/layers/cores/conv2d.py	36.66%	15 Missing and 4 partials ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##            main    #217      +/-   ##
========================================
+ Coverage   7.35%   7.38%   +0.02%     
========================================
  Files         58      58              
  Lines       5939    5947       +8     
  Branches    1005    1006       +1     
========================================
+ Hits         437     439       +2     
- Misses      5473    5478       +5     
- Partials      29      30       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[KonstantinWilleke]

Thanks a lot Max! I've built that branch with the sensorium model and tried it out - worked as intended.

[KonstantinWilleke]

The branch is out of date with the base branch though - could you update it? then I'll go ahead and merge

[MaxFBurg]

Thank you @KonstantinWilleke ! Merged the latest changes from main, so this PR should now be ready for merging into main.