Make the cube root to be inlined

[kimikage]

This makes the cube root to be inlined to make it suitable for vectorization.
This is intended to reduce precompilation problems rather than increase the speed. (cf. #425)

This also optimizes the pow5_12(::Float32). Edit: I will separate this into a new PR. (Edit2: see PR #493)

[codecov]

Codecov Report

Merging #489 (48109d3) into master (12f95f5) will increase coverage by 0.22%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #489      +/-   ##
==========================================
+ Coverage   93.12%   93.35%   +0.22%     
==========================================
  Files           9        9              
  Lines        1018     1053      +35     
==========================================
+ Hits          948      983      +35     
  Misses         70       70              

Impacted Files	Coverage Δ
src/conversions.jl	99.68% <100.00%> (+<0.01%)	⬆️
src/utilities.jl	98.75% <100.00%> (+0.33%)	⬆️

---

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 12f95f5...48109d3. Read the comment docs.

[kimikage]

using Colors, BenchmarkTools
rgb_f64 = rand(RGB{Float64}, 1000, 1000);
xyz_f64 = XYZ{Float64}.(rgb_f64);
xyz_f32 = XYZ{Float32}.(xyz_f64);
lab_f64 = Lab{Float64}.(xyz_f64);
lab_f32 = Lab{Float32}.(lab_f64);

julia> @btime convert.(Lab, $xyz_f64);
  121.341 ms (5965520 allocations: 113.91 MiB) # before
  29.326 ms (2 allocations: 22.89 MiB) # before (recompiled)
  24.800 ms (2 allocations: 22.89 MiB) # after

julia> @btime convert.(Lab, $xyz_f32);
  129.864 ms (5965520 allocations: 102.47 MiB) # before
  23.362 ms (2 allocations: 11.44 MiB) # before (recompiled)
  23.656 ms (2 allocations: 11.44 MiB) # after

I'm still not sure what's going on. 😕