Quadruple test testproc_9 broken

[Juergen Reuter]

Pacify missing in phs_base_write for lorentz_transformation and vector$ p and q

[kilian]

Fixed (hopefully) in r5000 (celebrate?).

I couldn't actually check quadruple, but at least the numerical noise is gone from the double precision output. We now have a ?sample_pacify flag.