Auto center frequency in slices with `freq_order`

[RemingtonRohel]

Specifically, there is an error raised when attempting to use multiple frequencies in one slice that require re-tuning in order to accommodate. The quick fix is to not use freq_order, but it is convenient and desired for e.g. normalsound so it would be nice to have this work long-term.

My first thoughts are that internally the slice should determine center frequencies for each frequency in freq_order then re-tune as required before starting a new averaging period. Not sure about the scope of changes required to make this work.

[RemingtonRohel]

Possible idea: remove the rxctrfreq and txctrfreq keywords from allowed slice parameters, instead always opting to calculate them automatically. The actual value itself can be stored as part of AveragingPeriod, instead of being a field of ExperimentSlice.

If freq is specified as a list, then the center frequencies can also live as lists, matching the number of elements of freq. Care must be taken in ExperimentPrototype.set_center_frequencies() to use freq_order then so that re-tuning is done as infrequently as possible (i.e. try to set center frequencies so that adjacently-used frequencies use the same center freq).