much better fix for avoiding seg faults for mass mesh sampling, SHOUL…

…D cause mass mesh sampling to no longer fail to produce a dataset. See commit 351517f, 351517f.

mass mesh sampling is still very slow, slower even than the momentum solver, which is annoying for mass mesh vtk and turbulence outputs, but at least this should cause it to run more consistently

src/ninja/ninjafoam.cpp

@@ -2427,7 +2427,7 @@ void NinjaFoam::writeProbeSampleFile( const wn_3dArray& x, const wn_3dArray& y,
         fprintf(fout, "        type    points;\n");
     }
     fprintf(fout, "        axis    xyz;\n");
-    fprintf(fout, "        ordered yes;\n");
+    fprintf(fout, "        ordered no;\n");
     fprintf(fout, "        points  $points;\n");
     fprintf(fout, "    }\n");
     fprintf(fout, ");\n");