Physical constants could be more precise

[joaoabcoelho]

I see some differences between Prob3++ and OscProb due to mismatches in physical constants. It would be nice to collect the constants in a more central place for easy checking and also make them more precise with some references on where the values came from.

For example G_F differs from OscProb by 0.04%:

Prob3plusplus/mosc.c

Line 26 in 02cf494

static const double tworttwoGf = 1.52588e-4;

1.5264932e-4 in OscProb

While the LoEfac differs by 0.005%:

Prob3plusplus/mosc.c

Line 489 in 02cf494

const double LoEfac = 2.534;

2.533865379 in OscProb

While these are quite small differences, they can accumulate when L/E is large.