Docs / UX: what's the difference between "queued" and "pending" messages?

## Summary
The CLI surfaces two distinct concepts — "queued" messages and "pending" messages — without explaining how they differ. From the user's perspective they look like the same thing (a message you've typed but the agent hasn't processed yet).

## Questions to resolve
- Is there a meaningful behavioral difference (lifecycle, when each is flushed, when each can be cancelled)?
- If yes: please document it in `/help` and in the on-screen labels (e.g. tooltip on the queue indicator).
- If no: please unify the terminology to a single word in the UI.

## Why this matters
Existing issues about queued-message editing/cancellation (#2055, #2378, #2905, #1857, #3692) all use "queued"; some other surfaces use "pending." Users can't tell whether they're filing duplicate reports about the same thing or two genuinely different bugs.

Start by comparing the CLI's uses of "queued" and "pending," then inspect the /help text and the on-screen queue indicator labels mentioned in the issue. Determine whether the terms describe different lifecycles, flushing behavior, or cancellation states; document the distinction in /help and the tooltip, or standardize the terminology if they are equivalent.

Written by the indexing model from the issue text.