Store text documents and CRDT state in file-system

[adzialocha]

We want to persist multiple things per document:

1. Document hash (for p2p networking)
2. The text itself
3. State-based CRDT / "snapshot" data inside of p2panda operation

There's different ways to approach it, but here's a rough idea:

1. Users can "Import" any text document into Aardvark, they will not have a document hash yet, another option is to simply just create a "New" one
2. In the moment they share it with other peers (through actively going through the "Share" flow) they get an document hash assigned
3. By default this document gets now stored in a hidden place with some markdown frontmatter holding the extra information to allow p2p networking:

   ---
   document_id: "7aa0c9a6e17e4eae7e5d490d83c0136d6047ffb4d37e201f3c1e57d198e4bcc7"
   ---
   
   # My text document
   
   Hello!

4. CRDT data is stored in an SQLite database next to these markdown documents, we simply map the document_id to the CRDT snapshot (only one required per document)
5. For now we can have an "Export" option which would allow users to save the document without the frontmatter. They'll loose the shared, collaborative meta-data for it and would need to re-share it again if they decided to re-import it again