add custom feature extractor

[hspitzer]

implements a custom feature extractor, that allows the user can define a function to extract features.
This allows easy adoption of e.g. DL feature extractors, but is more general than that, because every function can be used to extract features. A test is implemented.
What do you think?

I'm happy to hold of merging this after the ImageContainer API improvements are done, @michalk8 - whatever you think works best.

[codecov]

Codecov Report

Merging #245 (bccb033) into master (3bbb5f3) will decrease coverage by 0.02%.
The diff coverage is 75.00%.

@@            Coverage Diff             @@
##           master     #245      +/-   ##
==========================================
- Coverage   78.41%   78.39%   -0.03%     
==========================================
  Files          30       30              
  Lines        2775     2786      +11     
  Branches      447      450       +3     
==========================================
+ Hits         2176     2184       +8     
- Misses        458      460       +2     
- Partials      141      142       +1     

Impacted Files	Coverage Δ
squidpy/im/features.py	75.00% <0.00%> (-3.27%)	⬇️
squidpy/im/feature_mixin.py	87.35% <88.88%> (+0.01%)	⬆️
squidpy/_constants/_constants.py	100.00% <100.00%> (ø)

[michalk8]

Thanks @hspitzer !

I will merge this, but in #239, I suggest the following change:

• feature_fn should be always mandatory - this gets rid of this assertion
• feature_name should be optional (e.g. feature_fn.__name__ if present, otherwise e.g. custom_feature)

[hspitzer]

Thanks @hspitzer !

I will merge this, but in #239, I suggest the following change:

• feature_fn should be always mandatory - this gets rid of this assertion
• feature_name should be optional (e.g. feature_fn.__name__ if present, otherwise e.g. custom_feature)

great! Isn't feature_name already optional? I think it has a default value, no? But good idea to get it from feature_fn directly!