* feat: add LLM-friendly documentation and agent skill

Add llms.txt (concise overview) and llms-full.txt (full method reference)
following the llmstxt.org standard, plus a cross-client Agent Skill
(.agents/skills/) compatible with 30+ coding agents and a Claude Code
plugin for marketplace distribution.

- generate_llms_txt.py: auto-generates both files by introspecting Client
- .agents/skills/python-binance/: agentskills.io-compatible skill
- python-binance-plugin/: Claude Code plugin wrapping the same skill
- README.rst: add LLM & AI Agent Support section with npx install command

* refactor: remove plugin, compact llms-full.txt

- Remove python-binance-plugin/ directory
- Switch llms-full.txt to compact format: one heading + description +
  params per method, no response examples (360KB -> 198KB)
- Drop RST-to-Markdown conversion code (no longer needed)
- Truncate long param descriptions to 80 chars

* refactor: slim llms-full.txt to ~18K tokens

Switch to compact one-line-per-method format with param names only.
Truncate descriptions to 120 chars, drop auto-generated placeholder
descriptions. Down from 48K to 18K tokens.

* fix: resolve all ruff and pyright errors

- client.py: fix extra space after keyword, spacing around colon
- helpers.py: add missing space after comma
- code-generator.py: fix operator spacing, indentation issues
- examples/depth_cache_threaded_example.py: fix indentation, add type ignores

* fix: ignore E203 in ruff config (conflicts with ruff-format slice style)