Llama-cpp-http

Added:
- Homepage URL.
- Repository URL.

Removed:
- Removed `pyopencl` package from requirements.

Added:
- HTTP Server based on `aiohttp`, and deployed using `gunicorn`.
- HTTP Client for LangChain based on `websockets` for sync calls.
- `pyopencl` is used to determin available **OpenCL** devices (GPUs).
- Instructions in `LLAMA_CPP.md` how to build [llama.cpp](https://github.com/ggerganov/llama.cpp) w/o accelaration.
- Cahcing using `PonyORM` and `sqlite3`.