Add an HMISynopticMap documentation note that we are assuming that CDELT1 is the wrong sign

[ayshih]

We currently strip the sign from CDELT1 for HMI synoptic maps because they appear to have the incorrect sign. See #4053 (comment) for relevant discussion. However, we neglected in that PR to add a note to the docstring. Let's add a note.

---

Original issue text:

Per discussion on Element, the metadata for HMI synoptic maps is inconsistent with how they are (officially?) plotted:

• The metadata has negative CDELT1, which would means that the longitude axis should be plotted decreasing from 360 deg to 0 deg from left to right
• The JSOC-hosted PNGs of the same files show the longitude axis increasing from 0 deg to 360 deg from left to right (e.g., http://jsoc.stanford.edu/data/hmi/synoptic/hmi.Synoptic_Mr.2191.png)

Our code currently (unintentionally?) strips the sign from CDELT1 by calling np.abs() here:

sunpy/sunpy/map/sources/sdo.py

Lines 182 to 183 in f10e1dd

	return SpatialPair(np.abs(self.meta['cdelt1']) * self.spatial_units[0] / u.pixel,
	180 / np.pi * self.meta['cdelt2'] * u.deg / u.pixel)

That is, our output matches the JSOC-hosted PNGs, but if this is the "correct" way to plot synoptic maps, we should add a note to the docstring saying that we are intentionally not following CDELT1 as specified in the metadata.

[ayshih]

Thanks, @alasdairwilson, for reminding me that I already investigated this (#4053 (comment)), and I concluded that the CDELT1 in the metadata is actually wrong (i.e., provides the incorrect mapping between data and coordinates). This is not merely an issue of a plotting convention.