StandardScaling with separate constructor and fit methods to replace MeanStdScaling

[mzgubic]

Closes #106, closes #59, closes #87

This is made non-breaking by deprecating the MeanStdScaling.

To do:

• mention fit! in the docs

[codecov]

Codecov Report

Merging #107 (fdbe463) into main (1c86cbd) will increase coverage by 0.14%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #107      +/-   ##
==========================================
+ Coverage   98.74%   98.88%   +0.14%     
==========================================
  Files          13       15       +2     
  Lines         159      179      +20     
==========================================
+ Hits          157      177      +20     
  Misses          2        2              

Impacted Files	Coverage Δ
src/FeatureTransforms.jl	100.00% <ø> (ø)
src/deprecated.jl	100.00% <100.00%> (ø)
src/fit.jl	100.00% <100.00%> (ø)
src/scaling.jl	95.65% <100.00%> (+1.53%)	⬆️
src/apply.jl	100.00% <0.00%> (ø)
src/periodic.jl	100.00% <0.00%> (ø)
src/one_hot_encoding.jl	100.00% <0.00%> (ø)

---

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 1c86cbd...fdbe463. Read the comment docs.

[glennmoy]

In general I'm in favour of these changes - I just don't think we should extend StatsBase without good reason, and I'd prefer go with a Union type in the constructor.

Oh and we should also document fit! as part of the API and expect stateful transforms to have a method for it as well as apply.

[mzgubic]

Thank you both for reviewing, I think it's ready to go now. We went with the mutable approach so as not to inflate the API with an extra Fit class.