Bug fix for parentheses getting dropped from cell definitions.

[MicahGale]

Description

This fixes a bug where parentheses disappear from the geometry definition of a cell. This was because surfaces.HalfSpace essentially had no way to store that information, and when it was time to export the geometry would override the old parentheses.

This was solved in a few ways.

1. Introduce Operation.GROUP to replace Operation._SHIFT in this context. This was because this will be more visible to users and GROUP is clearer. Also _SHIFT is used for certain leaf nodes in complex geometry, and lead to confusion.
2. Adding an unbalanced tree level to HalfSpace to represent GROUP
3. Updating HalfSapce to ensure that GROUP is properly handled on export.
4. Made a method to detect when a group needs to be added for scratch geometries. This is done whenever a union node is below an intersection node. This is the proper order of operations (see MontePy deletes parentheses from cell regions #490).

To sanity check this logic:

    *
   / \
  +   3
 /\
1  2

Should be (1 : 2) 3 not 1 : 2 3

Fixes #490

Checklist

• I have performed a self-review of my own code
• I have added tests that prove my fix is effective or that my feature works (if applicable)
• Implement unit testing.
• Clear out all TODO
• Look into disappearing comments with @tjlaboss

[MicahGale]

Should we test writing all files out to file at least once and see what changes?

[tjlaboss]

Should we test writing all files out to file at least once and see what changes?

Eventually, we should test writing out each type of input at least once. This can be done piecemeal, with file read=... coming last.

[MicahGale]

Should we test writing all files out to file at least once and see what changes?

Eventually, we should test writing out each type of input at least once. This can be done piecemeal, with file read=... coming last.

I think this should wait until IO.StringIO is supported.

We already bulk read all valid test inputs through the checker. There's already good machinery for finding all valid input files. I think doing this enmass would be a good idea. We just run the risk of discovering a lot of bugs all at once that might is out of scope for this work I think.

[tjlaboss]

NICE