Speed up Pulse Sequence code

[mwalsh161]

• the Chanel's get.offset requires filesystem IO because it uses getpref - this should be cached instead (similar to module prefs).

• the NIDAQ has lots of overhead in the tasks to help with debugging; might be nice to include a mode that ignores that overhead (fast mode) by removing listeners and setting all relevant states to an unknown value.

[mwalsh161]

Could also have an option to APD pulse sequence's start method to bypass compiling and loading steps. Would need to follow the stack up to the BuildPulseSequence method in the PulseSequenceSweep superclass.