update for msc and timer changes in libavr32

[samdoshi]

Please merge monome/libavr32#26 first

[ngwese]

I have some concerns about the change to dac_update_now.

The reason the interrupts are being disabled is that the dac update timer function is not re-entrant. The arp/midi code uses dac_update_now to force a dac change outside of the normal timer callback based slewing in order to minimize the latency between trigger and cv output changes for a note.

It would probably be better to replace the original interrupt masking with timer pause/resume instead of eliminating it entirely.

[samdoshi]

There is no change in dac_update_now, I've just deleted some commented out lines. Or am I getting confused by something?

[ngwese]

Ack. You are correct, I didn't realize they were commented out. Now I'm curious if that means the potential race condition I was thinking about is present (I should know given that the code came from me...)