Misc improvement ideas

[autonull]

Some ideas, based on patham9's fork.

• Add containerization support
  Introduce a Dockerfile (based on Arch Linux) along with a run.sh script that handles building, running, shell access, cleaning, and volume mounting. This makes the entire environment reproducible and isolated while allowing host-mounted source code and persistent memory.

• Create a flexible environment configuration system
  Add .env.example (and support for .env) to easily configure multiple LLM providers via environment variables. Pair this with container_run.sh to auto-detect the provider and generate the appropriate provider_init.metta file at startup.

• Integrate LiteLLM for unified LLM access
  Update the Python helper (lib_llm_ext.py) to use LiteLLM. This removes hardcoded model logic and adds seamless support for OpenAI, Anthropic (Claude), Ollama, Groq, OpenRouter, and any other LiteLLM-compatible backend.

• Enhance observability and logging
  Introduce agent_run.py as a wrapper that filters noisy PeTTa output, supports a “dry-run” mode (for testing without calling LLMs), and writes structured JSON-line logs to agent.log with timestamps and event types.

• Strengthen memory management
  Update src/memory.metta to include configurable limits (maxHistory, maxRecallItems), time-based episode recall, and better response cleaning.

• Revised system prompt
  Also enhance the system prompt in memory/prompt.txt with clearer instructions for goal pursuit, memory usage, and error handling.

• Refine the core agent loop
  Modify src/loop.metta to use helper functions for parenthesis balancing, response cleaning, and smoother state transitions. This improves command parsing and overall stability of the interaction loop.

• Add proper project hygiene files
  Include a solid .gitignore (covering Python artifacts, .env, memory files, etc.) and a .dockerignore to keep Docker builds lean and prevent accidental leakage of secrets or cache files.

• Introduce version tracking and documentation overhaul
  Add a VERSION file and significantly expand README.md with quick-start guides, provider setup examples, command references, architecture notes, and local development instructions.

• Enable persistent memory with volume mounts
  Configure Docker volumes so the memory/ directory (including history.metta) is read-write mounted from the host. This preserves agent memory and learned skills across container restarts.