feat!: implement convolutional 2D layer

[c0dearm]

Convolutional layers with output dimensions inferred at compile time! This required a refactor of the project to be able to make it work, see: https://stackoverflow.com/questions/72742278/satisfying-a-trait-bound-with-a-const-generic-expression-is-it-possible

Unfortunately this requires nightly until the const equality and expressions are stabilized.

[codecov-commenter]

Codecov Report

Merging #17 (87fd5c6) into main (4287874) will decrease coverage by 2.86%.
The diff coverage is 77.27%.

@@            Coverage Diff             @@
##             main      #17      +/-   ##
==========================================
- Coverage   90.73%   87.86%   -2.87%     
==========================================
  Files          13       13              
  Lines         669      676       +7     
==========================================
- Hits          607      594      -13     
- Misses         62       82      +20     

Impacted Files	Coverage Δ
src/tensor/params.rs	50.00% <0.00%> (-19.24%)	⬇️
src/nn/losses.rs	81.81% <50.00%> (ø)
src/tensor/mod.rs	85.38% <57.14%> (-4.75%)	⬇️
src/nn/layers/dropout.rs	88.88% <77.77%> (-4.66%)	⬇️
src/lib.rs	97.22% <100.00%> (+5.33%)	⬆️
src/nn/activations.rs	91.30% <100.00%> (+0.12%)	⬆️
src/nn/layers/conv2d.rs	82.45% <100.00%> (-1.68%)	⬇️
src/nn/layers/linear.rs	93.75% <100.00%> (ø)
src/tensor/constant.rs	100.00% <100.00%> (ø)
src/tensor/variable.rs	97.50% <100.00%> (ø)
... and 3 more

---

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 4287874...87fd5c6. Read the comment docs.