Decouple database data from user data

[dustinvtran]

src/scrape_video_games.py currently writes the data scraped from video game databases into the user's JSON files. It's better to have a clean separation. Therefore it scales to >1 user and avoids issues such as users accidentally overwriting database data or database data accidentally overwriting user data.

[ghost]

Subtasks

• [x] Create new games, films, tvShows, books collections
  • [x] The fields keywords (see below) and apiRefs should be indexed
• [x] Populate those collections with existing entries' commonMetadata objects
  • [x] In order to potentially allow for searching in the future, the new documents should also contain a keywords field which contains an array of words (e.g. ["Inglourious", "Basterds", "Quentin", "Tarantino", "2009", "War", "Action", "Cristoph", "Waltz", "Brad", "Pitt"]). This is due to searching limitations in Fauna. This way, after setting up an index for that field, we can search any of those words to get that document. (partial word or fuzzy matching is not supported) I'm not going to bother with this for now because this can easily be tacked on at a later date, and I don't understand yet the best strategies to achieve good search. However, it seems very possible to achieve partial and fuzzy search, see this answer => https://stackoverflow.com/a/62131536/13791710
• [x] Existing entries' commonMetadata objects should be replaced with workRef: "refIdOfTheNewlyCreatedDocument"
  • [x] for now I'll just leave commonMetadata to not break the site while this is in progress. We can delete this field later, or leave it there. Future entries won't contain commonMetadata.
• [x] The "Retrieve work" endpoint should retrieve from our own database if exists (for example, we can query by index apiRefs, looking for the value { "name": "igdb", "ref": "1234" }), if not, then it should first create the database document (connecting to any necessary external APIs to get the data), then return it
• [x] When validating an "add entry" / "edit entry", instead of commonMetadata we just need to use the ref to the work's document
• [x] The "Retrieve user's entries" endpoint, before returning the entries to the client, should replace the references to Work documents with a commonMetadata object

[ghost]

https://github.com/td-memo/memo/commit/44ee8936f3367d7f7ccea035dc7851a589eb06da

Some bugs might come up, let's stay cautious.