Remove warnings when resampling=Holdout(rng=...)

[ablaom]

In light of JuliaAI/MLJBase.jl#559, this PR removes warnings added to mitigate #12, which did not really resolve the issue, but which the recent MLJBase PR renders moot.

Specifically, resampling can now be nothing, any Holdout, or any vector of the form [(train, test), ], where train and test are vectors of valid row indices for the data, and there will be no "cold restart" issues.

[codecov-commenter]

Codecov Report

Merging #22 (3519a76) into dev (af9c484) will increase coverage by 0.10%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##              dev      #22      +/-   ##
==========================================
+ Coverage   95.29%   95.40%   +0.10%     
==========================================
  Files           6        6              
  Lines         170      174       +4     
==========================================
+ Hits          162      166       +4     
  Misses          8        8              

Impacted Files	Coverage Δ
src/constructors.jl	96.66% <100.00%> (+0.51%)	⬆️

---

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 af9c484...3519a76. Read the comment docs.