fix: handle already existing collection.json

[ajacombs]

If the collection_from_items.py is run with a --uri source of an S3 path which already contains a collection.json file which is the output of a previous run of the script, it will fail.

This is because of a KeyError exception for "collection", which all the item STAC files have, but which the resulting collection.json file does not, as the collection ID is instead stored in the "id" key.

Swapping from item_stac["collection"] to item_stac.get("collection") eliminates this KeyError exception, and allows the script to run happily with a previously existing collection.json file, as it will be skipped and then overwritten.