Make VideoStream aware of px_per_unit

[asinghvi17]

Description

Fixes #3961 by making VideoStream aware of px_per_unit settings. This assumes that:

• All backends have px_per_unit (maybe this could be a get instead)

Type of change

Delete options that do not apply:

• Bug fix (non-breaking change which fixes an issue)

Checklist

• Added an entry in CHANGELOG.md (for new features and breaking changes)
• Added unit tests for new algorithms, conversion methods, etc.

[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	5.07s (5.00, 5.14) 0.05+-	111.56ms (109.35, 114.61) 1.84+-	415.34ms (412.32, 425.96) 4.83+-	9.40ms (9.25, 9.59) 0.11+-	25.89ms (25.75, 26.08) 0.13+-
master	5.07s (5.03, 5.11) 0.03+-	110.78ms (109.25, 111.53) 0.74+-	415.27ms (412.13, 418.73) 2.58+-	9.29ms (8.41, 9.58) 0.39+-	25.96ms (25.78, 26.24) 0.17+-
evaluation	1.00x invariant, -0.0s (-0.08d, 0.88p, 0.04std)	0.99x invariant, 0.78ms (0.56d, 0.33p, 1.29std)	1.00x invariant, 0.07ms (0.02d, 0.97p, 3.71std)	0.99x invariant, 0.11ms (0.37d, 0.51p, 0.25std)	1.00x invariant, -0.07ms (-0.46d, 0.41p, 0.15std)
CairoMakie	4.73s (4.54, 4.83) 0.10+-	106.18ms (101.34, 109.65) 3.51+-	163.33ms (154.83, 173.12) 6.19+-	9.21ms (8.94, 9.53) 0.22+-	1.16ms (1.07, 1.22) 0.06+-
master	4.74s (4.53, 4.85) 0.14+-	106.83ms (100.83, 109.74) 3.60+-	162.57ms (152.66, 169.34) 6.75+-	9.24ms (8.78, 9.55) 0.26+-	1.13ms (1.08, 1.21) 0.06+-
evaluation	1.00x invariant, -0.01s (-0.11d, 0.85p, 0.12std)	1.01x invariant, -0.65ms (-0.18d, 0.74p, 3.55std)	1.00x invariant, 0.76ms (0.12d, 0.83p, 6.47std)	1.00x invariant, -0.03ms (-0.12d, 0.82p, 0.24std)	0.98x invariant, 0.02ms (0.42d, 0.44p, 0.06std)
WGLMakie	5.61s (5.29, 5.86) 0.22+-	122.33ms (109.73, 158.06) 16.84+-	5.77s (5.40, 6.27) 0.32+-	13.11ms (11.59, 14.05) 0.88+-	116.95ms (113.64, 119.86) 2.28+-
master	5.90s (5.61, 6.41) 0.33+-	117.81ms (110.21, 125.80) 6.40+-	5.32s (4.99, 5.98) 0.37+-	12.10ms (11.97, 12.40) 0.17+-	120.98ms (116.37, 128.72) 3.77+-
evaluation	1.05x noisy🤷‍♀️, -0.29s (-1.04d, 0.08p, 0.27std)	0.96x invariant, 4.51ms (0.35d, 0.53p, 11.62std)	0.92x slower❌, 0.45s (1.30d, 0.03p, 0.34std)	0.92x slower❌, 1.0ms (1.58d, 0.02p, 0.53std)	1.03x faster ✓, -4.03ms (-1.29d, 0.04p, 3.03std)

[SimonDanisch]

I think we should have a cross backend framebuffer_size(screen), that should be used here.
GLMakie already has that, which I extended to screen here:
#4485