fix: show Extra Usage Utilization before the month's first charge - #668

What

Extra Usage Utilization now shows 0.0% (and an empty bar) when extra usage is enabled but nothing has been charged yet this month. Before, it rendered nothing at all.

Why

Until the first extra-usage charge of the month, the usage API reports utilization: null while still reporting the monthly limit and the amount spent ($0):

"extra_usage": { "is_enabled": true, "monthly_limit": 5000, "used_credits": 0.0, "utilization": null, "currency": "USD" }

The widget treated the missing utilization as no data and rendered nothing. Extra Usage Used and Extra Usage Remaining showed $0.00 and the full limit next to it, so the empty slot looked broken on every account with extra usage turned on. This shows up on a Max account right after turning extra usage on, and the Enterprise response in usage-fetch.test.ts, which mirrors a real account, has the same shape.

How

When the API reports no utilization, the widget works it out from the amount spent and the monthly limit (spent ÷ limit × 100), the same ratio the API reports once there's a charge. The API's own value still wins whenever it's present. Without a monthly limit (or with a limit of 0) the widget renders nothing, as before (#413). No extra fetches: the spent and limit fields come in the same response and are already cached, and the existing rule that treats missing extra-usage detail fields as conclusive is unchanged.

Demo

The same line rendered from a fake usage cache (no login, no API call): first at $0 spent with no utilization from the API, then at $12.50 of a $50 limit, which is unchanged.

Powerline: before

Powerline: after

Plain: before

Plain: after

Testing

• New tests: the utilization is derived at $0 and at $12.50 of $50; the API value wins over the derived one; no limit or a 0 limit still renders nothing.
• On today's main (with docs: explain an empty status line in one folder (workspace trust) #606 and feat(git-is-fork): render forks as an editable glyph #617 merged in locally): bun test gives 2785 pass, 0 fail (Bun 1.4.2), and bun run lint passes.
• The changed test file under Node (Vitest): 12 pass (9 existing and 3 new).
• Built CLI under Bun 1.4.2 and Node 26.10.0 with a usage cache at $0 spent: all runtimes give identical output in plain and Powerline modes, and it differs from main only by the added Utilization output. The TUI opens and exits cleanly under both.
• Piped render from a usage cache at $0 spent: main renders no Utilization widgets, this branch renders Overage: 0.0%. With an API-reported utilization the output is unchanged.