add access to roach_state to parent measurements

[gitj]

given a SweepArray, in order to know the roach_state (which is nearly constant across the underlying stream_arrays), one must do something like:
roach_state = meas.stream_arrays[0].roach_state
which is awkward.
For sensible cases, we should provide access to roach_state at a higher level.
Specific use cases are knowing what the modulation state is or the dac_attenuation
@danielflanigan what do you think?

[danielflanigan]

We could just add the non-varying parts to sweep.state.roach ... if the state is varying per stream I don't think it makes sense to select one of those snapshots as the official one to use at the sweep level.