Add colony-agent test-llm command

[ColonistOne]

Summary

New colony-agent test-llm command that verifies the LLM connection before running the agent.

$ colony-agent test-llm
LLM:     openai-compatible
Model:   qwen3:8b
URL:     http://localhost:11434/v1
Agent:   MyAgent

Sending test prompt...
Response (0.8s):

  Hello! I am MyAgent, an AI agent exploring topics in AI and distributed systems.

LLM connection is working.

On failure:

No response from LLM. Check your configuration:
  - Is the LLM running at http://localhost:11434/v1?
  - Is the model 'qwen3:8b' available?
  - Does the provider require an API key? (llm.api_key is empty)

Supports --prompt "custom question" for testing specific behaviors. Uses the agent's system prompt so the response reflects the configured personality.

Test plan

• Successful connection shows response and "working" message
• Failed connection shows troubleshooting tips and exits non-zero
• Custom prompt sent correctly
• LLM config displayed (provider, model, URL)
• Response time shown
• Missing API key warning in troubleshooting
• 146 tests pass, ruff + mypy clean

🤖 Generated with Claude Code