refactor: skill-first architecture with LLM backend abstraction

docs: document pip install openpraxis in openclaw-knowledge-coach skill

ci: add PyPI packaging and GitHub Actions publish workflow

chore: rename openclawn to openclaw across skills and docs

feat: support OpenClawn skills workflow for local knowledge mastery

feat: support image inputs via vision API

docs: update README description and add Vision section

Update config, CLI and tests

feat(cli): add provider selection flags for multi-llm runtime

feat(llm): add multi-provider structured output compatibility

style(cli): improve rich terminal visuals without logic changes

feat: multi-turn practice with coach (PracticeMessage, CoachReply, coach_turn/human_turn, MAX_PRACTICE_ROUNDS=3)

Bug fixes, new CLI commands, DB helpers, prompt enhancements, 56 tests

chore: init project structure