SUPG stabilization

[agzimmerman]

Currently peclet solves the Standard Galerkin form without stabilization. High Peclet Number (Pe) problems are solved via refining the grid near sharp boundary layers (i.e. we artificially reduce the local Pe). This grid refinement would not be required if we instead solve the SUPG stabilized form.

It will be interesting to revisit the 1D high Pe example from Donea & Huerta's book: https://github.com/alexanderzimmerman/peclet/blob/master/tests/Donea_Huerta_5p17_Pe5.prm

Also add new MMS (method of manufactured solution) tests (both 1D and 2D) at high Pe, e.g. in 2D increase the maximum velocity for this test: https://github.com/alexanderzimmerman/peclet/blob/master/tests/MMS_2D_VariableVelocity.prm