test: Add testing the artifact graph when there's an execution error

[jtran]

Follow-up to #5068 to test artifact graph output when there's a KCL execution error. The frontend expects that we always generate it, even when there's an error.

Interestingly, the output is different for a cube when there's an error afterwards. Compare the generated output files of the old cube test vs. the new cube_with_error test. It's missing the outgoing solid3d_get_opposite_edge and solid3d_get_next_adjacent_edge commands and everything that results from them.

[qa-wolf]

QA Wolf here! As you write new code it's important that your test coverage is keeping up.
Click here to request test coverage for this PR!

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Updated (UTC)
modeling-app	✅ Ready (Inspect)	Visit Preview	Jan 31, 2025 11:25pm

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 85.95%. Comparing base (e0ef10e) to head (5c4c045).
Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #5154   +/-   ##
=======================================
  Coverage   85.95%   85.95%           
=======================================
  Files          90       90           
  Lines       32893    32906   +13     
=======================================
+ Hits        28272    28285   +13     
  Misses       4621     4621           

Flag	Coverage Δ
wasm-lib	85.95% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[franknoirot]

Cool tests and code consolidation. That is interesting about the getOppositeEdge; is that because those are "post-extrude" commands? I guess I don't understand why they are interrupted by the later error, if that is what happens

[jtran]

Correct. In do_post_extrude(), the engine messages are enqueued to the batch since the execution doesn't need the result. In the successful case, they're eventually flushed and sent to the engine. But when there's an error, nothing flushes.

As a consumer of the artifact graph, this is unintuitive. But I'm not sure if we should change it.

I'm pretty sure that it has always behaved this way, even in TS. Now it's revealed, thanks to our new tests.