meshCubatureNodes assumes simplices are affine

[tcew]

The meshCubatureNodesTri2D and Tet3D functions use linear maps to construct the physical cubature node coordinates.

If the WB physical nodes and WB2cubature interpolation matrix cubI are already constructed then building the physical cubature nodes is trivial

cubx = cubI*x; in matlab speak.

[noelchalmers]

This is probably part of the larger 'support curved simplex elements' project. There are several places right now, most notably the geo factors, where we're assuming affine elements.

[tcew]

There are decisions to be made about how to implement curvilinear simplices.

Do we use the same array for geofacs for both affine and non-affine elements with an inevitable offsets array.

Do we sort elements into curved and noncurved groups then just need a single offset.

Do we create separate meshes.

Do we intend to support curvilinear elements of differing degree in their physical map construction ( for which affine and isoparametric would just be two instances )?