Improve Relocatability for GLMakie and fix tests

[SimonDanisch]

The tests didn't really test relocatability, since we only moved the package dir, but not the GLMakie source code.
This changes GLMakie to load all shaders on precompile, to make them part of the system image, much like what RelocatableFolder does it, but more explicitely (since we're caching + loading the shaders anyways).
Also deletes some dead code.

[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	5.33s (5.31, 5.35) 0.02+-	117.39ms (114.80, 121.65) 2.54+-	436.93ms (429.68, 462.43) 11.64+-	9.83ms (9.69, 10.07) 0.14+-	26.61ms (26.46, 26.78) 0.11+-
master	5.28s (5.21, 5.39) 0.06+-	118.03ms (110.17, 124.32) 4.73+-	435.35ms (419.56, 450.43) 13.13+-	9.68ms (8.82, 10.18) 0.42+-	26.62ms (26.50, 26.74) 0.09+-
evaluation	0.99x invariant, 0.04s (0.89d, 0.14p, 0.04std)	1.01x invariant, -0.65ms (-0.17d, 0.76p, 3.63std)	1.00x invariant, 1.59ms (0.13d, 0.82p, 12.39std)	0.99x invariant, 0.15ms (0.46d, 0.42p, 0.28std)	1.00x invariant, -0.01ms (-0.06d, 0.91p, 0.10std)
CairoMakie	4.91s (4.87, 4.96) 0.04+-	110.64ms (109.62, 111.86) 0.74+-	167.45ms (166.29, 168.86) 0.92+-	9.65ms (9.36, 9.74) 0.14+-	1.21ms (1.14, 1.26) 0.03+-
master	4.90s (4.86, 4.95) 0.03+-	111.26ms (109.49, 113.84) 1.58+-	168.47ms (165.82, 172.66) 2.35+-	9.62ms (9.31, 9.74) 0.14+-	1.21ms (1.15, 1.23) 0.03+-
evaluation	1.00x invariant, 0.01s (0.37d, 0.50p, 0.03std)	1.01x invariant, -0.63ms (-0.51d, 0.37p, 1.16std)	1.01x invariant, -1.02ms (-0.57d, 0.31p, 1.63std)	1.00x invariant, 0.03ms (0.18d, 0.74p, 0.14std)	1.00x invariant, 0.0ms (0.03d, 0.95p, 0.03std)
WGLMakie	5.81s (5.53, 5.99) 0.15+-	131.25ms (113.25, 153.22) 15.06+-	5.92s (5.78, 6.09) 0.11+-	14.09ms (12.76, 14.96) 0.81+-	121.27ms (117.70, 123.96) 2.01+-
master	5.70s (5.55, 5.90) 0.13+-	118.74ms (111.45, 137.30) 8.89+-	5.69s (5.51, 5.78) 0.11+-	13.86ms (12.78, 16.06) 1.09+-	128.02ms (118.04, 169.25) 18.57+-
evaluation	0.98x invariant, 0.1s (0.73d, 0.20p, 0.14std)	0.90x noisy🤷‍♀️, 12.51ms (1.01d, 0.09p, 11.97std)	0.96x slower X, 0.23s (2.18d, 0.00p, 0.11std)	0.98x invariant, 0.23ms (0.24d, 0.66p, 0.95std)	1.06x noisy🤷‍♀️, -6.75ms (-0.51d, 0.38p, 10.29std)

[bjarthur]

+100 for all improvements to relocatability! there is also a similar outstanding PR for WGLMakie. still haven't found the time to finish that up.