Improvements for Measure types in the plot_aes

[Mattriks]

• I've added and/or updated the unit tests
• I've run the regression tests
• I've built the docs and confirmed these changes don't cause new errors
• Move 2 functions (currently in this PR) to Measures.jl
• I've squash'ed or fixup'ed junk commits with git-rebase

This PR:

• Fixes The function apply_statistic_typed doesn't handle type Measure #1357
• Provides better support for Measure types (from Measures.jl)
  (e.g. when the same aesthetic in a different layer has a different type)
• Paves the way for fixing other issues e.g. see example 2 here, which I leave to another PR

Example

# Examples from #1357
p1 = plot(xmin=[.25], xmax=[.75], Geom.band, color=[colorant"red"])
p2 = plot(xmin=[.25], xmax=[.75], Geom.band, color=[colorant"red"],
 layer(Geom.rect, xmin=[0], xmax=[1], ymin=[0], ymax=[1], color=[colorant"blue"]))
p3 = plot(x=[2,4], y=[2,4], size=[1.414], color=[colorant"gold"],
    Coord.cartesian(fixed=true) )
hstack(p1, p2, p3)

Note the y-axis on plot p1: the Measure units appear because those are the only units provided (cf. plot p2).

[codecov-io]

Codecov Report

Merging #1358 into master will decrease coverage by 0.07%.
The diff coverage is 95.45%.

@@            Coverage Diff             @@
##           master    #1358      +/-   ##
==========================================
- Coverage    90.6%   90.53%   -0.08%     
==========================================
  Files          38       38              
  Lines        3961     3950      -11     
==========================================
- Hits         3589     3576      -13     
- Misses        372      374       +2

Impacted Files	Coverage Δ
src/coord.jl	91.57% <100%> (+0.18%)	⬆️
src/statistics.jl	96.93% <100%> (+0.04%)	⬆️
src/aesthetics.jl	81.51% <66.66%> (-0.39%)	⬇️
src/ticks.jl	90.32% <0%> (-1.49%)	⬇️

---

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 bb210ab...74f1310. Read the comment docs.

[tlnagy]

I tagged a new version of Measures, which should be out soon with your changes.

ref: JuliaRegistries/General#6638

[Mattriks]

Thanks! I ran the regression testing and found 2 diffs. I show the devel-output here.
i) The first one is because the Measure units now appear on the respective axes (as expected):

p1 = evalfile(dirname(pathof(Gadfly))*"/../test/testscripts/hvband.jl")

ii) The second one (below) is unexpected (but correct). Here Gadfly turns the categorical sizes into integers 1:3, so the {x,y} aesthetic ranges get expanded by ±3. There are current bugs in the way the size aesthetic is handled. e.g. Scale.size_continuous(minvalue= , maxvalue= ) should rescale the size aesthetic to 0-1. Similarly there needs to be a better design outline for how discrete sizes should be treated. I leave that to another PR.

p2 = evalfile(dirname(pathof(Gadfly))*"/../test/testscripts/point_size_categorical.jl")

[Mattriks]

This is ready to merge, if there are no other comments.