NdArray bump follow up

[bytesnake]

This is the followup to the PR #110. It tries to simplify our story of cauchy::Scalar and num_trait::Float by moving the trait bound of cauchy behind an associated type and avoiding thus the ambiguity when mixing both bounds.

It adds a function Float::map_cauchy_bound and pulls in the trait bound here for an example in linfa-clustering: https://github.com/rust-ml/linfa/pull/121/files#diff-cc618ed2646ccdd41e994796498d4b9ed256dc0d99c1976302eba1c496caaf64R256.

I also added a F::cast function to shorten number casting and avoid num_traits::FromPrimitive 😅

@relf

[relf]

Maybe we could add num_traits::MulAdd<Output = F> (see sparsemat/sprs#278 (comment)) to ease the upgrade to sprs 0.10?

[bytesnake]

yes! Haven't seen the post

[codecov-commenter]

Codecov Report

Merging #121 (5f9eb25) into master (8276bdc) will decrease coverage by 0.07%.
The diff coverage is 39.52%.

@@            Coverage Diff             @@
##           master     #121      +/-   ##
==========================================
- Coverage   58.30%   58.23%   -0.08%     
==========================================
  Files          75       76       +1     
  Lines        6697     6700       +3     
==========================================
- Hits         3905     3902       -3     
- Misses       2792     2798       +6     

Impacted Files	Coverage Δ
...infa-clustering/src/appx_dbscan/hyperparameters.rs	18.51% <0.00%> (-3.71%)	⬇️
...hms/linfa-clustering/src/dbscan/hyperparameters.rs	6.25% <0.00%> (ø)
...clustering/src/gaussian_mixture/hyperparameters.rs	34.92% <0.00%> (ø)
...ms/linfa-clustering/src/k_means/hyperparameters.rs	40.90% <0.00%> (ø)
algorithms/linfa-clustering/src/k_means/init.rs	87.23% <ø> (-0.40%)	⬇️
algorithms/linfa-elasticnet/src/hyperparameters.rs	12.50% <0.00%> (ø)
algorithms/linfa-kernel/src/lib.rs	71.18% <0.00%> (+0.72%)	⬆️
algorithms/linfa-linear/src/ols.rs	71.96% <0.00%> (ø)
algorithms/linfa-pls/src/lib.rs	90.00% <ø> (ø)
algorithms/linfa-pls/src/pls_svd.rs	42.10% <ø> (ø)
... and 43 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 8276bdc...5f9eb25. Read the comment docs.