VisibilityXds: scan_number

[kettenis]

Using numbers to identify scans is probably workable; after all MSv2 uses numbers as well. However, VEX schedules used for VLBI use a string. Most astronomy observations are scheduled with scan labels that contain a number (i.e. "No0035") but geodetic VLBI uses strings that are basically - (i.e "134-1944"). So if there is an efficient way to assign scan labels instead of just numbers that would probably make it easier to trace back scans to the original schedule.

[Jan-Willem]

The overall philosophy of MSv4 is to be as descriptive as possible (using spw_names instead of spw_ids). I agree that we should consider moving from scan_numbers -> scan_names.

On a related note, I don't think this applies to baseline_id since it is unindexed dimension and the (baseline_antenna1_name, baseline_antenna2_name) should be used which are of length baseline_id.