Feature Request: Add Moorcheh as vector store backend for agent LTM

[Neelpatel1604]

Summary

Add Moorcheh as a new vector store backend for SuperAGI agent long-term memory (LTM), following the existing VectorStore + VectorFactory pattern used by Redis, Pinecone, Qdrant, and Weaviate.

Motivation

• SuperAGI has no unified memory provider - LTM is wired via VectorStore subclasses and VectorFactory
• Moorcheh is a managed, text-first semantic search/memory service - fits the existing add_texts / get_matching_text flow without requiring self-hosted vector DB infra
• Useful for teams that want serverless agent memory without running Redis/Pinecone/Qdrant

Proposed scope (v1)

• Add MOORCHEH to VectorStoreType enum
• New superagi/vector_store/moorcheh.py implementing VectorStore
• Wire into VectorFactory.get_vector_storage() and build_vector_storage()
• Config: MOORCHEH_API_KEY in config_template.yaml
• Add moorcheh-sdk to requirements.txt
• Unit tests under tests/unit_tests/vector_store/
• Allow LTM when LTM_DB=Moorcheh even for non-OpenAI models (Moorcheh handles embeddings)

Out of scope (follow-up PRs)

• Resource uploads via LlamaVectorStoreFactory
• Knowledge marketplace /connect/moorcheh endpoint
• GUI LTM picker update
• Per-agent LTM_DB wiring (stored in config but not used at runtime today)

Integration points

Layer	Files
Write	superagi/agent/output_handler.py
Read	superagi/tools/tool_response_query_manager.py
Bootstrap	superagi/jobs/agent_executor.py

Acceptance criteria

• LTM_DB=Moorcheh creates memory via VectorFactory
• Tool outputs are written to Moorcheh after each step
• ToolResponseQueryManager can recall relevant past tool output
• get_matching_text returns {"documents": [...], "search_res": "..."}
• Unit tests pass in CI
  ]