Implement polar_to_cartesian

[kimikage]

This adds sincos360, the rough equivalent of sincosd.
The sincos360 is less accurate than sincosd, but in many cases more accurate than sincos(deg2rad(x)).
Also, while sincosd(Inf) throws a DomainError, sincos360(Inf) returns (NaN, NaN).

This is mainly intended to mitigate precompilation and inlining issues. (cf. issue #496)

This might have to be moved to ColorTypes.jl as well as atan360. (cf. PR #495) Fortunately, moving those methods to ColorTypes.jl seems to maintain the benefits. However, I'm not sure why the problem occurs with Base.

[codecov]

Codecov Report

Merging #510 (957b257) into master (12549a8) will increase coverage by 0.06%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #510      +/-   ##
==========================================
+ Coverage   95.38%   95.45%   +0.06%     
==========================================
  Files           9        9              
  Lines        1171     1188      +17     
==========================================
+ Hits         1117     1134      +17     
  Misses         54       54              

Impacted Files	Coverage Δ
src/algorithms.jl	74.75% <100.00%> (-0.25%)	⬇️
src/conversions.jl	99.38% <100.00%> (ø)
src/utilities.jl	98.46% <100.00%> (+0.11%)	⬆️

---

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 12549a8...957b257. Read the comment docs.

[kimikage]

julia> lchab32 = [LCHab(100*rand(Float32), 100*rand(Float32), 360*rand(Float32)) for y = 1:1000, x = 1:1000];

julia> lchab64 = [LCHab(100*rand(Float64), 100*rand(Float64), 360*rand(Float64)) for y = 1:1000, x = 1:1000];

julia> @btime convert.(Lab{Float32}, $lchab32);
  30.570 ms (2000002 allocations: 41.96 MiB) # before
  12.473 ms (2 allocations: 11.44 MiB) # before (recompiled)
  8.589 ms (2 allocations: 11.44 MiB)  # after

julia> @btime convert.(Lab{Float64}, $lchab64);
  30.995 ms (2000002 allocations: 68.66 MiB) # before
  16.649 ms (2 allocations: 22.89 MiB) # before (recompiled)
  10.462 ms (2 allocations: 22.89 MiB) # after

[kimikage]

This has almost no impact on functionality (i.e. public APIs), so I will merge this in order to proceed with other tasks.