Floating Point Error due to small mesh cells

[chrisb2244]

Solver crashes when trying to 'correctPhi'. Likely the same problem occurs with a variety of solvers/fields.
Interpolation scheme causes a division by the volume of the cell (I think) and this is being rounded to zero after several layers of refinement.

[chrisb2244]

If memory serves, the hexRef8.C which serves as the model for hexRef4 contains handling for small faces - this might be to protect from this issue. Will take a look at the handling there and see if it is applicable to 2D. If so, not sure why it's not in hexRef4 already, but perhaps an oversight and overzealous removing of code.