MeshUtils.F90: update to calculate mesh pieces

[richb2k]

This is a followup post related to this prior post from the 3D Inductor thread:

https://www.elmerfem.org/forum/viewtopic.php?p=31254#p31254

There have been cases in the forum recently where 'Calculate Mesh Pieces' doesn't properly report the number of pieces in the mesh.

To help illustrate the issue, open the attached archive that contains an elmergrid input file, a case.sif, and a short script. The case.sif calls for the calculation of the mesh pieces and the calculation reports that only one exists. Examining the resulting mesh files in ElmerGUI show that there are actually two cubes, not connected to each other. Since the cubes don't share any nodes, this is a non-conforming mesh and the calculation should show two mesh pieces exist. The case.sif then continues and runs the Heat Equation solver, where the case.vtu results show that heat can't conduct through both cubes, since the two cubes are not connected.

The attached archive can also be run where the calculation will properly report the number of pieces, by editing the elmergrid input file and commenting out the line 'Order Nodes = 0.0 0.0 -1.0'. Including the 'Order Nodes' command changes the numbering order of the elements, such that the pieces calculation no longer works properly.

cubesZ.zip

[richb2k]

The issue has been fixed with PR #591