No-op fallback for restrict_to_volume! postrender operation

[hughcars]

Note: This is based on top of #112 as a required predicate.

Changes as part of #112 significantly reduced the number of entities being rendered to the SolidModel before postrendering. One consequence of this is to reduce the time required for all postrendering operations, as they generally scale superlinearly with number of entities. Additionally, because mesh entities were by nature colocated with many detailed piece of a component, this resulted in a significant amount of nodes, segments and surfaces that were non-reconcilable without a fragment operation. By removing these entities, the "auto union" operation commonly used with negative or positive layers has been made significantly more powerful, as the union with entity removal reduces the number of total surfaces enormously.

This has two consequences: 1) it reduces the number of entities that need to fragment, improving the run time significantly, and 2) it removes constraints on the mesher associated with adhering to these "construction" entities which aren't actually relevant to the physical geometry, thereby improving mesh speed and quality.

This can be most cleanly seen in the QPU17 example, where comparing main against the results from this branch, when run on an M1 mac using a single thread for meshing.

Before:

Rendering to SolidModel: 1965.336371 seconds (110.03 M allocations: 7.415 GiB, 0.06% gc time, 0.98% compilation time: <1% of which was recompilation)
Generating 1D Mesh: 270.173185 seconds
Generating 2D Mesh: 761.339575 seconds
Generating 3D Mesh: 478.235420 seconds

and after:

Rendering to SolidModel: 538.681900 seconds (103.71 M allocations: 6.577 GiB, 0.14% gc time, 3.48% compilation time: <1% of which was recompilation)
Generating 1D Mesh: 11.668969 seconds (155.80 k allocations: 7.668 MiB, 0.45% compilation time)
Generating 2D Mesh: 9.354619 seconds
Generating 3D Mesh: 231.924238 seconds

Giving a 3.64x speed up in render, 23.15x in 1D meshing, 81.39x in 2D meshing and 2.06x in 3D meshing, for an overall speedup of 4.38x. The most dramatic improvements are in the 1D and 2D meshing, which it's hypothesized comes from a) the huge reduction in total entities, and b) the simplification of those entities that remain.

[codecov]

Codecov Report

❌ Patch coverage is 93.54839% with 2 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/solidmodels/render.jl	83.33%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!

[gpeairs]

These two PRs are pretty exciting. Just one comment, and I'll take another look if there are any new changes once you're done testing, but otherwise LGTM

[gpeairs]

Did you mean to use these in the loop below?

[hughcars]

Yes, must've been a rebase error. I was trying to force away a yellow type safety warning from some unions, but I don't think it helped. I'll check it again and get rid of the debris.