Use polys for axis3

[jkrumbiegel]

Currently, Axis3 doesn't work well with svg export because the meshes that are used for the side panels cause rasterization in Cairo. This breaks svg display for a lot of frontends, for example in VSCode:

This PR uses translated polys instead to avoid this.

[MakieBot]

Compile Times benchmark

Note, that these numbers may fluctuate on the CI servers, so take them with a grain of salt. All benchmark results are based on the mean time and negative percent mean faster than the base branch. Note, that GLMakie + WGLMakie run on an emulated GPU, so the runtime benchmark is much slower. Results are from running:

using_time = @ctime using Backend
# Compile time
create_time = @ctime fig = scatter(1:4; color=1:4, colormap=:turbo, markersize=20, visible=true)
display_time = @ctime Makie.colorbuffer(display(fig))
# Runtime
create_time = @benchmark fig = scatter(1:4; color=1:4, colormap=:turbo, markersize=20, visible=true)
display_time = @benchmark Makie.colorbuffer(fig)

	using	create	display	create	display
GLMakie	6.38s (6.29, 6.47) 0.07+-	167.24ms (163.77, 171.39) 2.80+-	477.35ms (471.17, 482.69) 4.53+-	8.32ms (8.08, 8.46) 0.13+-	26.34ms (26.22, 26.50) 0.10+-
master	6.50s (6.42, 6.55) 0.04+-	167.41ms (162.59, 172.38) 3.52+-	483.62ms (475.03, 496.46) 8.26+-	8.33ms (8.19, 8.52) 0.10+-	26.32ms (26.23, 26.43) 0.09+-
evaluation	1.02x faster ✓, -0.11s (-2.02d, 0.00p, 0.05std)	1.00x invariant, -0.17ms (-0.05d, 0.92p, 3.16std)	1.01x invariant, -6.26ms (-0.94d, 0.11p, 6.40std)	1.00x invariant, -0.0ms (-0.03d, 0.96p, 0.12std)	1.00x invariant, 0.02ms (0.18d, 0.74p, 0.10std)
CairoMakie	6.59s (6.48, 6.72) 0.10+-	198.38ms (187.36, 208.70) 7.18+-	227.00ms (214.94, 242.79) 9.11+-	9.25ms (8.89, 9.67) 0.24+-	1.35ms (1.32, 1.39) 0.02+-
master	6.60s (6.52, 6.68) 0.06+-	203.24ms (192.98, 217.55) 9.19+-	232.56ms (214.74, 248.68) 14.15+-	9.25ms (8.92, 9.54) 0.24+-	1.33ms (1.30, 1.39) 0.03+-
evaluation	1.00x invariant, -0.01s (-0.15d, 0.78p, 0.08std)	1.02x invariant, -4.86ms (-0.59d, 0.29p, 8.19std)	1.02x invariant, -5.56ms (-0.47d, 0.40p, 11.63std)	1.00x invariant, 0.0ms (0.00d, 0.99p, 0.24std)	0.98x invariant, 0.03ms (0.95d, 0.10p, 0.03std)
WGLMakie	6.68s (6.50, 6.91) 0.17+-	167.18ms (160.28, 177.27) 5.91+-	5.62s (5.48, 5.76) 0.10+-	12.71ms (12.24, 13.38) 0.42+-	131.80ms (125.61, 139.26) 4.96+-
master	6.60s (6.54, 6.73) 0.07+-	165.58ms (161.60, 168.08) 2.17+-	5.87s (5.77, 6.00) 0.07+-	12.39ms (12.10, 12.67) 0.20+-	135.40ms (132.32, 139.88) 2.86+-
evaluation	0.99x invariant, 0.07s (0.55d, 0.33p, 0.12std)	0.99x invariant, 1.6ms (0.36d, 0.52p, 4.04std)	1.05x faster ✓, -0.26s (-2.92d, 0.00p, 0.09std)	0.97x invariant, 0.32ms (0.97d, 0.11p, 0.31std)	1.03x invariant, -3.6ms (-0.89d, 0.13p, 3.91std)