fix #1546 for real (inserting newly created scenes)

[SimonDanisch]

fix for #1546, which was supposed to be fixed by #2428, but somehow didn't make it in there...

[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(display(fig))

	using	create	display	create	display
GLMakie	30.30s (30.19, 30.52) 0.11+-	16.45s (16.37, 16.56) 0.07+-	15.23s (15.13, 15.37) 0.09+-	10.68ms (10.54, 10.81) 0.10+-	115.79ms (114.65, 117.53) 0.94+-
master	30.36s (30.27, 30.47) 0.06+-	16.58s (16.46, 16.65) 0.07+-	15.24s (15.10, 15.39) 0.09+-	10.63ms (10.48, 10.76) 0.10+-	114.97ms (114.21, 116.15) 0.65+-
evaluation	-0.20%, -0.06s invariant (-0.70d, 0.22p, 0.09std)	-0.76%, -0.13s faster ✓ (-1.77d, 0.01p, 0.07std)	-0.09%, -0.01s invariant (-0.15d, 0.78p, 0.09std)	+0.43%, 0.05ms invariant (0.45d, 0.41p, 0.10std)	+0.71%, 0.82ms invariant (1.01d, 0.09p, 0.79std)
CairoMakie	26.51s (26.38, 26.58) 0.06+-	16.36s (16.29, 16.44) 0.05+-	2.51s (2.50, 2.53) 0.01+-	10.58ms (10.53, 10.66) 0.05+-	4.20ms (4.11, 4.29) 0.08+-
master	26.46s (26.41, 26.53) 0.04+-	16.37s (16.31, 16.40) 0.04+-	2.51s (2.50, 2.56) 0.02+-	10.48ms (10.46, 10.52) 0.02+-	4.10ms (3.98, 4.20) 0.07+-
evaluation	+0.20%, 0.05s invariant (0.97d, 0.10p, 0.05std)	-0.05%, -0.01s invariant (-0.17d, 0.75p, 0.05std)	-0.00%, -0.0s invariant (-0.01d, 0.99p, 0.02std)	+0.95%, 0.1ms slower X (2.61d, 0.00p, 0.04std)	+2.35%, 0.1ms slower X (1.36d, 0.03p, 0.07std)
WGLMakie	36.56s (36.45, 36.65) 0.08+-	20.22s (20.12, 20.35) 0.08+-	24.11s (23.45, 24.38) 0.32+-	15.18ms (14.58, 15.96) 0.52+-	2.42s (2.19, 2.54) 0.11+-
master	36.73s (36.54, 37.17) 0.22+-	20.32s (20.03, 20.59) 0.23+-	24.12s (23.38, 24.53) 0.38+-	14.98ms (14.40, 15.97) 0.54+-	2.44s (2.37, 2.54) 0.07+-
evaluation	-0.47%, -0.17s invariant (-1.02d, 0.10p, 0.15std)	-0.48%, -0.1s invariant (-0.57d, 0.32p, 0.15std)	-0.07%, -0.02s invariant (-0.05d, 0.93p, 0.35std)	+1.32%, 0.2ms invariant (0.38d, 0.49p, 0.53std)	-1.20%, -0.03s invariant (-0.31d, 0.58p, 0.09std)

[SimonDanisch]

I don't think the regression/improvements can be from this PR... Will merge this, but just to be sure I'll open a test pr!