[CRITICAL] Provide Full Rate Limit Transparency

[laz-001]

Problem

the rate-limit chaos, see:

https://github.com/search?q=repo%3Agithub%2Fcopilot-cli+rate+limit&type=issues

Proposed solution

Besides the discussion here:

• Improve rate limit UI for all types of rate limit #2827

Some more things:

• => extended to per model family rate limit display (e.g. gpt/claude)
• => the agents should be able to access the actual rate limits (and see the notifications), so they can adapt.
• => a prompt or subagent call can contain a rate limit (we can set a limit ourselves, for each call)

Those are must-have constructs in order to be able to control workflow behavior decently.

Example prompts or workflows

Do this and that.

/token-limit: n

Additional context

META: (this form) this really needs to be a multiline input: