Allow absolute (ms) control of position/window

[rconstanzo]

In addition to the relative/phase-based pos/win controls allow them to also be set with the alternate message types detailed in #16 (position 356.25 ms, window 4410 samples, etc...).

If any set messages are sent (either set or setloop) then the pos/win remain fixed/absolute. (The current behavior keeps the relative/phase positions which generally changes the absolute size of pos/win.

There are some funky edge cases though.
Like changing (via set) from a large to a small buffer where the previous absolute window size is larger than the new overall loop length. In this circumstance the window size would be clamped to the loop size.
However, things get interesting when then switching back to a large loop size. If possible [karma~] should keep track of the precious absolute window size and return to it, unclamping the new window size.