Instead of treating the polytrope as a free boundary problem I have defined an interpolating polynominal, accurate to within 0.01 percent over n=[0,5) which is used to set the size of the domain for a given n
The NewtonSolver has been subclassed to try to auto enforce the zero boundary central condition by modifying the residual vector and the gradient matrix. This is a work in progress
BREAKING CHANGE:
Rescaling a mesh by a linear transformation is a useful option so that we can start with a single "base" mesh and then rescale it to the dimensions needed for our problem. This commit adds the LinearRescale option too meshIO so that a unit sphere can be turned into a sphere of arbitrary radius (as an example).
previously I had a lagrangian multipliers at every element; however, we are enforcing a global constraint so there need only be one lagrangian multiplier
