docs: align deployment and command docs with runtime behavior

[eeee0717]

Summary

• align internal command examples with actual built-in commands (skill instead of removed tools/skills.*)
• fix README config variable names to match AgentSettings (BUB_MAX_*, BUB_MODEL_TIMEOUT_SECONDS)
• fix Docker deployment docs for .agents mount path and clarify BUB_HOME bind-mount caveat
• expand deployment env check command to include provider-scoped BUB_API_KEY/BUB_API_BASE
• add missing skill validation limits and legacy .agent/skills note
• fix README docs links to point to docs/ paths

[PsiACE]

Thanks