x86-64: Simplify and optimize vdso clock_gettime monotonic variants

We used to store the wall-to-monotonic offset and the realtime base.
It's faster to precompute the monotonic base.

This is about a 3% speedup on Sandy Bridge for CLOCK_MONOTONIC.
It's much more impressive for CLOCK_MONOTONIC_COARSE.

Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: John Stultz <john.stultz@linaro.org>