Preparing for rules ingestion

Reducing intro. Adding arguments about DND acting as an abstract knowledge domain. Removed 'configuration' section. Added a bit to the 'chat' section

Ignoring internal folders

Adding retrieval context to the chatmessage table

removing .vscode

Adding pytest

Adding some retrieval context to the chatmessage table. NOTE: This will break if source documents are updated, which they are intended to be. This may need to change. For example, instead of simply holding what ids are retrieved, we could change the column to JSONB and save the id, content, source doc, and even the similarity score

Making similarity optionnal to accomodate the way we retrieve source chunks from our database

Adding the 'lore-seed' context. Will be replaced with a database-native solution eventually

Expanding append_message to accomodate saving retrieved chunk ids

Accomodating the 'lore seed' which the user can configure to provide hard-coded context to each prompt

Accomodating the 'lore seed' which the user can configure to provide hard-coded context to each prompt

ensuring we render chunks that are returned from a chat

Fixing crazy rendering bug

Couple important updates. First, enabling the access of retrieved lore chunks from our database, for each chat we have. Second, adding an API section for the lore seed we use to start each converstaion. We can now look at it and update it. This will be replaced with a database-native approach eventually. Enabling chunk retrieval in our 'chat' endpoint