Also do not assign a double to an integer

Rather than

  (1-wc)*Ma + wc*Cv

we now have

  Ma + Cv

Also change its default from 1e6 to 1e5 and treat it as size_t

The solvers for u0/a0 and v now each have parameters tolerance,
maximumIterations, verbosity

Also, do a 1-6-0 TNNMG with (six) V-3-3 cycles

The code for Laursen's model is long gone

This reverts commit 17bd6ae8e51c2cbdcf9ec87bbcb79c87ab81c61d.