WIP: expand data by FrequencyWeights

[joshday]

This PR adds a repeat(x::RealVector, w::FrequencyWeight{<:Integer}) method to create a new vector where each element in x gets repeated by its associated weight in w. Would this be useful, if only for better testing? It would help catch cases like #313

x = randn(10)
w = rand(1:5, 10)
p = [0, .25, .5, .75, 1]
@test quantile(x, w, p) == quantile(repeat(x, w), p)

[codecov]

Codecov Report

Merging #314 into master will decrease coverage by 0.18%.
The diff coverage is 0%.

@@            Coverage Diff            @@
##           master    #314      +/-   ##
=========================================
- Coverage   90.88%   90.7%   -0.19%     
=========================================
  Files          18      18              
  Lines        1964    1957       -7     
=========================================
- Hits         1785    1775      -10     
- Misses        179     182       +3

Impacted Files	Coverage Δ
src/weights.jl	86.09% <0%> (-2.8%)	⬇️
src/hist.jl	95.38% <0%> (-0.07%)	⬇️
src/moments.jl	99.4% <0%> (-0.01%)	⬇️
src/cov.jl	100% <0%> (ø)	⬆️
src/sampling.jl	92.11% <0%> (+1.03%)	⬆️

---

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 7e9d724...588fc45. Read the comment docs.

[joshday]

Oops, this already exists.

StatsBase.inverse_rle([1,2,3], fweights([1,2,3]))

[nalimilan]

It would make sense to rename that function as it's pretty hard to discover. I even had filed an issue against Julia to add this kind of function without noticing it existed here: JuliaLang/julia#16443.