Fix Issue #1015

Proposal for a fix for #1015 (closed).

I removed the iterative algorithm for solving the generalized Kepler equation (CartesianOrbit.meanToEccentric). Switching to the use of the perifocal frame analagously to how it is done for the hyperbolic case allows the use of the Gooding-Odell algorithm in KeplerianAnomalyUtility for solving the standard Kepler equation.

I had to make very minor bumps to a bunch of tolerances, particularly in the GNSS attitude tests.

Note: EquinoctialOrbit and CircularOrbit also have their own algorithms for solving the generalized Kepler equation which might suffer from a similar numerical issue. I haven't tried to do anything with those classes.