Refactored Symbolic Neural Networks

[benedict-96]

Completely changed the interface of SymbolicNeuralNetworks.

The relevant objects are:

1. SymbolicNeuralNetwork
2. Jacobian
3. Gradient
4. build_nn_function

All these functions/structs have well-documented docstrings. In addition there use is also documented in the readme and in the markdown docs.

[codecov]

Codecov Report

Attention: Patch coverage is 71.57534% with 83 lines in your changes missing coverage. Please review.

Project coverage is 72.30%. Comparing base (e05b6d8) to head (e09ee57).
Report is 74 commits behind head on main.

Files with missing lines	Patch %	Lines
src/hamiltonian.jl	0.00%	28 Missing ⚠️
src/custom_equation.jl	0.00%	25 Missing ⚠️
src/pullback.jl	64.00%	9 Missing ⚠️
src/symbolic_neuralnet.jl	57.89%	8 Missing ⚠️
src/layers/dense.jl	50.00%	3 Missing ⚠️
src/utils/build_function.jl	93.87%	3 Missing ⚠️
src/utils/build_function2.jl	91.66%	3 Missing ⚠️
src/utils/latexraw.jl	0.00%	2 Missing ⚠️
src/derivatives/derivative.jl	94.11%	1 Missing ⚠️
src/layers/abstract.jl	0.00%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main      #16       +/-   ##
===========================================
+ Coverage   47.61%   72.30%   +24.68%     
===========================================
  Files          13       16        +3     
  Lines         252      325       +73     
===========================================
+ Hits          120      235      +115     
+ Misses        132       90       -42     

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

[benedict-96]

I just saw that the Documentation test is failing for now. This is because the package manager is still installing GeometricMachineLearning v0.3.5 instead of v0.3.6. I'll let the CI run again tomorrow ...