Two non-dimensional parameters are added: "dissipation_number" and "gruneisen"
under the Solver component. One can use the original incompressible solver by
setting "gruneisen=0". The code will treat this as "gruneisen=infinity". 
Setting non-zero value to "gruneisen" will switch to compressible solver.

One can use the TALA solver for incompressible case by setting "gruneisen" to
a non-zero value while setting "dissipation_number=0". This is useful when
debugging the compressible solver.

Two implementations are available: one by Wei Leng (U. Colorado) and one by
Eh Tan (CIG). Leng's version uses the original conjugate gradient method for
the Uzawa iteration and moves the contribution of compressibility to the RHS,
similar to the method of Ita and King, JGR, 1994. Tan's version uses the
bi-conjugate gradient stablized method for the Uzawa iteration, similar to the
method of Tan and Gurnis, JGR, 2007. Both versions agree very well. In the
benchmark case, 33x33x33 nodes per cap, Di/gamma=1.0, Ra=1.0, delta function
of load at the mid mantle, the peak velocity differs by only 0.007%. Leng's
version is enabled by default. Edit function solve_Ahat_p_fhat() in
lib/Stokes_flow_Incomp.c to switch to Tan's version.