Intent extractor (getzep#101)

* Intent Extractor

* Updated, cleaned up intent_extractor + test

* remove duplicate search test

* Moved the goroutine into Extract, keeping Notify pretty simple

* refactor processing of message intents into separate function to increase readability. refactor error handling to catch all errors.

* group extractor config models

* fix merging issue with metadata maps

* check storeMetadataByPath return value

* better warning for .env file missing

---------

Co-authored-by: Daniel Chalef <[email protected]>

increase makeEmbedRequest and docker-compose health timeouts (getzep#93)

* increase makeEmbedRequest context timeout to 10s
* increase NLP docker-compose health timeouts

Implement Local Embeddings using SentenceTransformers (getzep#92)

* local embedding models

* ENABLE_EMBEDDINGS=true

hybrid search (getzep#85)

Implement alongside vector search:

- hybrid search over metadata using jsonpath queries.
- hybrid search using message creation timestamps

cloud_start to address render config issues (getzep#79)

EntityExtractor + Metadata (getzep#78)

- nlp service container
- EntityExtractor + tests
- Render blueprint updated for nlp service

OpenAI Retry Client (getzep#77)

* OpenaiRetryClient wrapper

Summarizer not summarizing all messages with large batch message puts (…

…getzep#72)

* refactor GetMemory: simpilify and return all if no summary

* fix summarizer issue with retrieving all msgs. refactor GetMessages

* break up pg operations in order to increase maintainability

Allow zep to start without config file (getzep#61)

* allow zep to run without config file

add retry and timeout to openai calls (getzep#49)

* add retry and timeout to openai completion calls

* retry for embeddings

* upgrade retry to v4