Guide.shapekey updates

[Mattriks]

• I've updated the documentation to reflect these changes
• 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

This PR:

• closes making scatter plots with different shaped markers (issues) #1366 (see Roadmap #1385 Roadmap)
• fixes triggering of shapekey (issue 1 in making scatter plots with different shaped markers (issues) #1366, plot p1 below)
• fixes issue 6 in making scatter plots with different shaped markers (issues) #1366 (plot p2)
• shapekey integration with 1-length size and color vectors (plot p3)

Example

x, y = 1:5, rand(5)
p1 = plot(x=x, y=y, shape=[Shape.square], size=[8pt])

p2 = plot(Theme(point_size=6pt, alphas=[0.6]),
    layer(x=0.5:4, y=rand(4), shape=["Gr1"]),
     layer(x=4*rand(4), y=rand(4), shape=["Gr2"]) )

p3 = plot(x=x, y=y, shape=x, size=[6pt], color=[colorant"orange"])
hstack(p1,p2,p3)

[codecov-io]

Codecov Report

Merging #1411 into master will decrease coverage by 0.72%.
The diff coverage is 92.00%.

@@            Coverage Diff             @@
##           master    #1411      +/-   ##
==========================================
- Coverage   90.31%   89.58%   -0.73%     
==========================================
  Files          39       39              
  Lines        4242     4291      +49     
==========================================
+ Hits         3831     3844      +13     
- Misses        411      447      +36     

Impacted Files	Coverage Δ
src/guide/keys.jl	83.06% <84.61%> (-0.97%)	⬇️
src/misc.jl	65.06% <100.00%> (-2.46%)	⬇️
src/scale.jl	97.55% <100.00%> (+0.08%)	⬆️
src/theme.jl	65.07% <0.00%> (-4.42%)	⬇️
src/color_misc.jl	60.60% <0.00%> (-3.92%)	⬇️
src/Gadfly.jl	76.66% <0.00%> (-2.45%)	⬇️
src/aesthetics.jl	80.00% <0.00%> (-1.31%)	⬇️
src/bincount.jl	94.89% <0.00%> (-0.98%)	⬇️
src/data.jl	0.00% <0.00%> (ø)
... and 2 more

---

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 3a2e115...a96839e. Read the comment docs.

[Mattriks]

The regression tests showed the following changes. With this PR, in testscripts:

1. "point_color_shape_size.jl", "point_shape_explicit.jl", and "shape_hvline.jl" the shapekey is not triggered (this is correct). Note that the examples in the first two testscripts are written in "old Gadfly style" (pre-shapekey) and could be updated in the future.

2. "key_columns.jl" shows that the size aesthetic will override any default settings in Theme. This maybe an issue in a plot that uses multiple aesthetics (e.g. distinct size and shape mappings). Need to think more about plot p3 above (and if necessary add more to this PR).

[tlnagy]

This is great! I'm very excited for the propagation of a single color and shape through a layer. Makes multiple layer plots a lot easier. How hard would it be to autowrap non-array values? So you could do something like color=colorant"orange" instead of color=[colorant"orange"], which makes more sense from a user's perspective.

[Mattriks]

The syntax color=colorant"orange" needs investigation (not in this PR). Unfortunately for this PR, I've rollbacked the changes that made plot p3 possible, because a more elaborate design is really needed to make that work properly (e.g. for 2 layers with different keys). But this PR still closes #1366 (plot p1 and p2 above). I should regression test again.

[Mattriks]

The regression tests now show changes to just the 3 testscripts listed in point 1 in my regression test post above. With this PR, in those 3 testscripts the shapekey doesn't trigger with shape=[Shape.function]. This is consistent with e.g.color=[colorant"color"] which doesn't trigger the colorkey.

This PR is now ready to merge.

[bjarthur]

is it worth putting this in the gallery?

[Mattriks]

Done. On Travis, for some reason the Codecov bit failed on Julia 1 just now.
Worked after reopening!

[Mattriks]

This is ready to merge.