Another update for cut atom

[tristan-huber]

Updates logic around cut operations in abundance.

Specifically it turns out that Volume computation for 3d solids is fast. Fast enough that we can use it to double check the class of interaction which a cut operation falls into. The cases being:

1. cut between two shapes that don't touch -> volume of both inputs is unchanged
2. cut which entirely removes the target shape -> volume of result is 0
3. cut removes some but not all of the target -> volume of result is < initial but > 0

Only the third case requires a new entry in the cache.

Prior to this PR case 1 was fairly common and would lead us to re-instantiate identical objects. This was particularly pervasive in assemblies. In an initial effort to fix this I conflated cases 1 and 2 which would sometimes allow shapes to persist after when they should have been cut away to nothing.

[tristan-huber]

TODO:

• update fuse and intersect to use the booleanOperation helper. Intersect can sometimes produce no-ops and sometimes produce empty shapes. Fuse can sometimes produce no-ops.
• add instrumentation to booleanOperation helper so that it also logs into our metrics like createIfAbsent does