Avoid deepcopy'ing Shape.topo_parent.

[fischman]

Speeds up benchy example from 27s to 5.5s.

Motivated by #909 adding examples that take a while to run.
Discussion in Discord: thread.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.93%. Comparing base (bda0a6a) to head (4d8dfe1).
Report is 4 commits behind head on dev.

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #914      +/-   ##
==========================================
+ Coverage   96.73%   96.93%   +0.19%     
==========================================
  Files          32       32              
  Lines        9434     9459      +25     
==========================================
+ Hits         9126     9169      +43     
+ Misses        308      290      -18     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[gumyr]

I'm going to try this out...

[fischman]

I'm going to try this out...

Any findings?
(no rush, just curious)

[gumyr]

My current project uses selectors extensively so it should exercise this code thoroughly. Mainly I'm selecting faces which aren't as numerous as the vertices of the benchy example so I'm seeing only modest improvements in performance - say 308 seconds to 297 seconds for a long test. Functionality seems to be fine but I've seen some things I can't explain so more investigation is required.