Add option to retrieve ivar with get_valid_data methods

[dkirkby]

The get_valid_data methods from SpecFile, PlateFile and FrameFile all return dflux calculated from internal ivar values. This is often what you want, but sometimes ivar is more useful, e.g., when averaging spectra. I propose to keep dflux as the default, but add an option to get ivar instead of (or in addition to?) dflux.

[dmargala]

Similarly, sometimes loglam is useful (instead of wavelength) for compatibility with already written software.

[dkirkby]

Yes, good point. That should clearly be an either/or option. Can you think of a use case for returning both ivar and dflux?

[dcunning11235]

One case where both are used, though getting ivar would be sufficient: averaging together fluxes (ivar as weights) and also outputting the combined sigma/dflux.

[dkirkby]

I am working on the averaging use case right now to try and figure this out...

[dkirkby]

Closed with PR #88