1D wcs transformations not returning single array

The documentation for wcs transformations state that the all_world2pix, etc, operations return a single array if the input is a single array. However, currently, they return a list of length 1 with the single array of pixel coordinates (in the case of all_world2pix/world_to_pixel):

In [6]: my_wcs = fitswcs.WCS(header={'CDELT1': 1, 'CRVAL1': 6562.8, 'CUNIT1': 'Angstrom', 'CTYPE1': 'WAVE', 'RESTFRQ': 1400000000, 'CRPIX1': 25})

In [7]: my_wcs.world_to_pixel(np.arange(10) * u.nm)
Out[7]:
[array([-6538.8, -6528.8, -6518.8, -6508.8, -6498.8, -6488.8, -6478.8,
        -6468.8, -6458.8, -6448.8])]

This is also inconsistent with the current GWCS interface that returns a single array of coordinates.

[astrofrog]

Relating to some of the discussion above, I’m not interested in changing the default behavior of the ‘native’ methods for GWCS and FITS-WCS, just the APE 14 API - which I think is still fine to do since it was new in 3.1.

[astrofrog]

Thinking about this some more, I’d be ok with returning variables not in a list/tuple in the single coordinate case, but this means that chaining as:

wcs2.pixel_to_world(*wcs1.world_to_pixel(...))

won’t work. One way around this is to say that pixel_to_world and world_to_pixel should be able to accept a tuple, and in this case, it should auto-expand it, so that we could then do:

wcs2.pixel_to_world(wcs1.world_to_pixel(...))