Preserve accessories

[Ariana-B]

See #168
Assume that any assets that are not bands are accessories, and make sure they are captured in the Dataset metadata doc

[github-actions]

🚀 Deployed on https://f6d43600145741f0--odc-stac-docs.netlify.app

[alexgleith]

Looks good to me! I'm happy to approve, but I think we should wait for Kirill before we merge.

[Ariana-B]

@Kirill888 could you please review

[Kirill888]

Why start full then remove?
Let's start empty and add as we go instead

[Ariana-B]

Well I can't think of a better way of determining which assets are accessories other than by discounting the ones we know to be bands. Instead of _acc_names I could define a list of band_names and then do a set difference, but I don't think that's a cleaner solution than the current one.

[Kirill888]

• remove is O(n), repeated in a O(n) loop, so this is O(n^2) operation after this change

I'd rather this was a set operation, this also makes it easier to read to me, as it would be a single line that fully defines what _acc_names is, rather than currently where you start with "all asset names" then need to search code below to understand that this then becomes "all asset names except for data assets". Also if asset_name in _acc_names: will always be true in this logic.

[Ariana-B]

That's fair; updated.

[Kirill888]

@Ariana-B can you please add code to strip function

odc-stac/odc/stac/model.py

Lines 394 to 398 in b46d41c

	def strip(self) -> "ParsedItem":
	"""
	Copy of self but with stripped bands.
	"""
	return replace(self, bands={k: band.strip() for k, band in self.bands.items()})

to set accessories to an empty dict

[Kirill888]

Thanks for this @Ariana-B