Checks: Perform normal calculation using full integration

[kopperp]

Previously, we use the cross-product of the first two tangent vectors for the normal calculation. This can produce wrong results for highly deformed elements which are nonetheless correct. Perform the full high-order integration on the Jacobian to avoid this issue.

[github-actions]

☂️ Code Coverage

current status: ✅

Overall Coverage

Statements	Covered	Coverage	Threshold	Status
3208	2925	91%	85%	🟢

New Files

No new covered files...

Modified Files

File	Coverage	Status
pyhope/basis/basis_watertight.py	100%	🟢
pyhope/io/io_debug.py	98%	🟢
pyhope/mesh/extrude/mesh_extrude.py	99%	🟢
pyhope/script/pyhope_cli.py	98%	🟢
TOTAL	99%	🟢

updated for commit: d31d5a9 by action🐍

[kopperp]

@scopplestone Technically, we are now checking this twice since CheckWatertightalso reports inward facing normals, but only for connected sides. Given that the check is expensive, I'm open for suggestions.

[scopplestone]

Any way to do the check only once for each side? Either extend the water tightness check by also checking the boundary sides or skip the connected sides in the new check?

[kopperp]

Any way to do the check only once for each side? Either extend the water tightness check by also checking the boundary sides or skip the connected sides in the new check?

Yes, we can just add the non-connected sides to the watertightness check. The only downside is that watertightness gets only checked after the MeshConnect, so we might try to connect wrongly-oriented sides and fail there.

Skipping connected sides doesn't work because we don't have that information at this point. The original idea was to have a cheap check right at the beginning and then only do the expensive verification right at the end.