Tweak mapping to improve plotting performance

[non-Jedi]

This was a learning exercise for me to practice using Julia profiling tools. I picked Gadfly's plot function as it seemed likely to have some low hanging fruit while being relatively simple to grok. These changes improve the benchmark posted below on my system from 18.7 microseconds to 10.3 microseconds through minor refactoring; definitely not earth-shattering but in the direction of goodness. I leave it to the maintainers discretion whether it's worth their time to review this patch.

Note that this only improves plot generation and not plot display, which takes the bulk of the time in an interactive plotting session.

julia> using Gadfly, BenchmarkTools

julia> x = 1:10000; y = rand(1:100, 10000); c = rand(1:250, 10000);

julia> @btime plot(x=$x, y=$y, color=$c);
  10.281 μs (39 allocations: 4.44 KiB)

Contributor checklist:

• I've updated the documentation to reflect these changes
• I've added an entry to NEWS.md
• I've added and/or updated the unit tests
• I've run the regression tests
• I've squash'ed or fixup'ed junk commits with git-rebase
• I've built the docs and confirmed these changes don't cause new errors

[codecov-io]

Codecov Report

Merging #1229 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1229      +/-   ##
==========================================
+ Coverage   83.53%   83.53%   +<.01%     
==========================================
  Files          35       35              
  Lines        4032     4033       +1     
==========================================
+ Hits         3368     3369       +1     
  Misses        664      664

Impacted Files	Coverage Δ
src/mapping.jl	83.96% <100%> (+0.62%)	⬆️
src/Gadfly.jl	75.72% <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 2dfff47...a70f656. Read the comment docs.

[bjarthur]

there's a nobel prize for whoever figures out the time-to-first problem. thanks for this!