feat: add active skills to slash commands, recursive skills folders and display options

[JamesDAdams]

Summary

• Active skills in slash commands (/): Enabled skills are now suggested in SlashAutocomplete across the chat composer (ChatInput) and TaskEditor with distinctive green styling (text-accent-success) and a Skill badge.
• Recursive skill directory discovery: Backend recursively traverses subdirectories inside skills folders to load nested SKILL.md packages, assigning the relative folder path as group.
• Folder grouping, token estimations & bulk toggle in Skills settings:
  • Skills located in subdirectories are organized into full-width collapsible cards (matching the MCP server cards style), collapsed by default.
  • Displays token count estimates for individual skills (~X tokens) and aggregated totals for skill folders.
  • Added a folder-level toggle to activate or deactivate all skills in that folder simultaneously without expanding/collapsing the accordion.
• Fullscreen slash command display option: Added a setting under Display ("Choose whether the commands view uses default sizing or fills the available screen height.") allowing the slash autocomplete menu to span the full screen height bounded 10px below the header.

AI-Enhanced Development

Tell what models helped shape this PR:

• AI Models: Gemini 3.7 Flash

Cache Impact

Does this PR affect anything cached — system prompts, tool definitions, skills, or other context?

• Yes — Recursive discovery allows nested skill packages to be discovered and exposed via /api/skills and available for dynamic skill loading (load_skill), with estimated token counts exposed in API responses.

[JamesDAdams]

[JamesDAdams]

@co-l I make the change for i18n translate