Client.get_collection should call STAC endpoint

[geospatial-jeff]

I think the pystac_client.Client.get_collection method should contact the GET /collections/{collectionId} endpoint instead of iterating through the response of GET /collections. Its better for an API client to utilize the server where appropriate instead of doing something client-side, and the later runs into issues for APIs which have a lot of collections that aren't necessarily exposed through the first call to GET /collections.

[matthewhanson]

@geospatial-jeff totally agree with this, the iteration seems unnecessary.

Just thinking this through....pystac-client works with both static catalogs and APIs. get_collection currently calls get_collections. In the case of an API it will call /collections. In the case of a static catalogs it will look at just the immediate children and find the right one. In a static catalog also, the link is not necessarily the collection ID so it has to be iterated through.

I think the right answer here is that get_collection should also check for conformance, and if this is an API that supports the /collections endpoint then it should use /collections/<cid>, otherwise resort to iterating through.

Will try and get this into the next minor release which should be in the next day.