[RFC] Add support for parsing 8-digit and 4-digit hex notations

[kimikage]

This adds the support for parsing 8-digit and 4-digit hex notations following the manner mentioned in #353 (comment).

Do not merge this PR before the decision of #353 for the consistency.

[codecov]

Codecov Report

Merging #371 into master will increase coverage by 0.1%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master     #371     +/-   ##
=========================================
+ Coverage   77.94%   78.04%   +0.1%     
=========================================
  Files          11       11             
  Lines         866      870      +4     
=========================================
+ Hits          675      679      +4     
  Misses        191      191

Impacted Files	Coverage Δ
src/parse.jl	97.01% <100%> (+0.18%)	⬆️

---

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 47e3750...e5f6923. Read the comment docs.

[kimikage]

Please note the following:

julia> parse(Colorant, "0xFF8800AA") # opaque purple
ARGB{N0f8}(0.533,0.0,0.667,1.0)

julia> parse(RGBA, "0xFF8800AA") # still opaque purple
RGBA{N0f8}(0.533,0.0,0.667,1.0)

We have another option, which refers to the first argument of parse(). However, I don't think it is good to depend on the context which doesn't appear in the string.

[timholy]

This is really nice. Do all those @inbounds annotations really make a difference?

[kimikage]

Perhaps there is little difference in terms of the speed on Julia v1.3.0 and v1.0.5. They are slower than v1.2.0 (especially on Windows) for some unclear reason.

I added @inbounds because the boundary checks are very disturbing when trying to clarify the cause of stall or the issue JuliaLang/julia#34055, rather than because of the speed. (In other words, the native codes are quite different.)

Since I don't have enough time to investigate and the cause has not been identified, @inbounds has actually little benefit for the end users for now.

[kimikage]

I merged #390. I also changed != nothing to !== nothing and removed some @inbounds.

[kimikage]

Since PR #387 was merged, we can merge this.