Add some precompiles

[timholy]

This adds some precompile statements to reduce latency. Because the methods are simple to infer, the savings are modest, but combined with precompiles in other packages (FixedPointNumbers, ColorTypes, ColorVectorSpace) I'm getting latency reductions of ~30% for some common operations. So these seem worth having.

[johnnychen94]

Would you mind to kindly point out where related documentation can be found? This topic looks interesting

[timholy]

https://github.com/timholy/SnoopCompile.jl

[kimikage]

Does precompiling have side effects?

julia> versioninfo()
Julia Version 1.3.0
Commit 46ce4d7933 (2019-11-26 06:09 UTC)
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: Intel(R) Core(TM) i7-8565U CPU @ 1.80GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.1 (ORCJIT, skylake)

master

julia> @time parse(RGB{Float32}, "#C0FFEE")
  0.233481 seconds (981.38 k allocations: 50.342 MiB, 4.99% gc time)
RGB{Float32}(0.7529412f0,1.0f0,0.93333334f0)

julia> @time parse(RGB{Float32}, "#C0FFEE")
  0.000017 seconds (13 allocations: 576 bytes)
RGB{Float32}(0.7529412f0,1.0f0,0.93333334f0)

julia> @time parse(Colorant, "#C0FFEE")
  0.001666 seconds (1.94 k allocations: 116.674 KiB)
RGB{N0f8}(0.753,1.0,0.933)

julia> @time parse(Colorant, "#C0FFEE")
  0.000010 seconds (11 allocations: 512 bytes)
RGB{N0f8}(0.753,1.0,0.933)

teh/precompile

julia> @time parse(RGB{Float32}, "#C0FFEE")
  0.098095 seconds (115.82 k allocations: 6.166 MiB)
RGB{Float32}(0.7529412f0,1.0f0,0.93333334f0)

julia> @time parse(RGB{Float32}, "#C0FFEE")
  0.000014 seconds (14 allocations: 592 bytes)
RGB{Float32}(0.7529412f0,1.0f0,0.93333334f0)

julia> @time parse(Colorant, "#C0FFEE")
  0.001681 seconds (1.94 k allocations: 116.439 KiB)
RGB{N0f8}(0.753,1.0,0.933)

julia> @time parse(Colorant, "#C0FFEE")
  0.000010 seconds (12 allocations: 528 bytes)
RGB{N0f8}(0.753,1.0,0.933)

Edit:
In this case, the difference is whether RGBA() is inlined or not. (Not again!)
I don't understand why this happens.

(This is a off-topic, but I should have added @inbounds for mat.captures[*]. I will do it when implementing the 4-digit/8-digit hex notation parser.)

[codecov]

Codecov Report

Merging #370 into master will increase coverage by 0.25%.
The diff coverage is 90%.

@@            Coverage Diff             @@
##           master     #370      +/-   ##
==========================================
+ Coverage   76.98%   77.24%   +0.25%     
==========================================
  Files          10       11       +1     
  Lines         817      835      +18     
==========================================
+ Hits          629      645      +16     
- Misses        188      190       +2

Impacted Files	Coverage Δ
src/Colors.jl	100% <ø> (ø)	⬆️
src/utilities.jl	94.73% <100%> (ø)	⬆️
src/precompile.jl	88.88% <88.88%> (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 1bc194e...eb734ea. Read the comment docs.

[kimikage]

master

julia> using Colors

julia> precompile(Tuple{typeof(parse),Type{ColorTypes.Colorant},String})
true

julia> @time parse(Colorant, "#C0FFEE")
  0.000055 seconds (12 allocations: 560 bytes)
RGB{N0f8}(0.753,1.0,0.933)

julia> @time parse(Colorant, "#C0FFEE")
  0.000008 seconds (11 allocations: 512 bytes)
RGB{N0f8}(0.753,1.0,0.933)

What's going on here?

Edit:
If there is a negative impact on performance, let's report the issue in each package and then merge.
JuliaMath/FixedPointNumbers.jl#148
JuliaGraphics/ColorTypes.jl#136
JuliaGraphics/ColorVectorSpace.jl#110