Fast trimmed least squares

[mapi1]

I had to add StatsBase as an dependency for sample. Not sure if this is the most performant implementation, profiling showed that most of the time is spent in sortperm (partialsortperm makes it slower somehow) and \ which seems reasonable. There exists a nested version for long data, which is currently not implemented. I hope the documentation is somewhat clear.

[codecov]

Codecov Report

Merging #18 (ad653ce) into master (43fc5b6) will decrease coverage by 1.09%.
The diff coverage is 92.45%.

@@            Coverage Diff             @@
##           master      #18      +/-   ##
==========================================
- Coverage   98.82%   97.73%   -1.10%     
==========================================
  Files           2        3       +1     
  Lines         256      309      +53     
==========================================
+ Hits          253      302      +49     
- Misses          3        7       +4     

Flag	Coverage Δ
unittests	97.73% <92.45%> (-1.10%)	⬇️

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

Impacted Files	Coverage Δ
src/TotalLeastSquares.jl	96.87% <ø> (ø)
src/flts.jl	92.45% <92.45%> (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 43fc5b6...ad653ce. Read the comment docs.

[baggepinnen]

Thanks for this contribution with tests and everything right from the get-go! I pushed a commit 7ad5e70 with some very minor changes (please ignore the erroneous correction of minimize, I must be tired :P), if you are okay with them, I'll go ahead and merge this PR :)

[mapi1]

Great! Your changes look good to me, small details that make the code much clearer. :)