[Feature Request] Expose per-model weekly rate limits in status line and add unattended session usage controls

Bug Description
▎ Max 20x user: 39% of my weekly limit (and ~54% of the Fable weekly limit) was consumed in under 24h by one unattended session that fanned out ~16 Fable max-effort reviewer subagents, each re-reading up to 300k tokens per call, and it ignored usage warnings injected by hooks. The status line JSON only exposes five_hour/seven_day, so the per-model weekly limit was invisible until it was too late. Support confirmed this is known Claude Code behavior and that nothing can be restored. Please: (1) expose per-model weekly limits in the status line rate_limits, (2) add a setting to cap subagent fan-out / effort in unattended runs, (3) let hooks enforce a usage budget reliably. Experiences like this make a paying 20x customer consider leaving.

Environment Info

• Platform: win32
• Terminal: vscode
• Version: 2.1.284
• Feedback ID: 46804dde-ecbd-412f-a228-59e398614fa5

Errors

[]