EOS max-mass fix and linear interp for k(m)

This is to fix the max-mass calculation in lalsimulation NS family code. It takes the last three points of an array, which includes a turnover, and tries to find the max. Then it replaces the last point in the array with the found max. However, I think it's possible to find a max point that comes before the second-to-last point and thus should replace the last two points.

Despite this fix, k2(m) still can go negative due to the cspline interpolation. Therefore, I also change the interpolation to linear for k2(m). Two things to note here: 1) the better option, I think, is to incorporate the steffen interpolation type, but this is not in the cluster versions of gsl; 2) I also don't think interpolating in logspace will help the spline for the particular corner case I've run into:

lalsim-ns-params -Q -q 32.827159 -1 4.339848 -2 4.099636 -3 2.144642 -m 1.523194