fix: refactor part id assignment post wrap

[mhoeijm]

Performance of the vtkImplicitPolyDataDistance filter is not great, but so far seems more robust than the vtkExtractEnclosedPoints filter.

• 44/44 Rodero/Strocchi cases mesh ok and give a valid heart_model.vtu
• Some cases failed on fiber generation, this seems resolved now
• Most following EP simulations seem ok in terms of validity of the .k files

[wenfengye]

is this covered by any test already?

[mhoeijm]

is this covered by any test already?

will do, it is in principle covered by test_model_generation but since this is not very strict in terms of mesh stats it will not capture changes in the methods very well (except if it really breaks methods downstream).