Pytorch interface autodifferentiation

[BrunoLiegiBastonLiegi]

This tries to adapt #20 to for incorporating autodifferentiation, both native with pytorch and custom by defining a new autograd.

[codecov]

Codecov Report

Attention: Patch coverage is 85.95890% with 41 lines in your changes missing coverage. Please review.

Project coverage is 62.22%. Comparing base (51037fd) to head (920889f).
Report is 30 commits behind head on model.

Files with missing lines	Patch %	Lines
src/qiboml/operations/differentiation.py	69.44%	22 Missing ⚠️
src/qiboml/models/keras.py	59.25%	11 Missing ⚠️
src/qiboml/models/decoding.py	95.94%	3 Missing ⚠️
src/qiboml/models/ansatze.py	83.33%	2 Missing ⚠️
src/qiboml/models/encoding.py	95.34%	2 Missing ⚠️
src/qiboml/models/pytorch.py	98.30%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##            model      #37      +/-   ##
==========================================
- Coverage   65.10%   62.22%   -2.88%     
==========================================
  Files          13       14       +1     
  Lines         682      826     +144     
==========================================
+ Hits          444      514      +70     
- Misses        238      312      +74     

Flag	Coverage Δ
unittests	62.22% <85.95%> (-2.88%)	⬇️

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

Files with missing lines	Coverage Δ
src/qiboml/backends/jax.py	46.00% <100.00%> (+2.84%)	⬆️
src/qiboml/models/pytorch.py	98.59% <98.30%> (-1.41%)	⬇️
src/qiboml/models/ansatze.py	85.71% <83.33%> (-14.29%)	⬇️
src/qiboml/models/encoding.py	95.34% <95.34%> (ø)
src/qiboml/models/decoding.py	95.94% <95.94%> (ø)
src/qiboml/models/keras.py	63.41% <59.25%> (-36.59%)	⬇️
src/qiboml/operations/differentiation.py	54.54% <69.44%> (ø)

... and 1 file with indirect coverage changes