fix(oauth): offer xAI chat models returned by GET /v1/models

[feeeeling]

Summary

• A signed-in xAI (Grok/X subscription) account now treats GET /v1/models as the list of conversation models it may use. grok-4.7 is offered as soon as Grok returns it, without waiting for the pinned pi-ai catalog to ship that id.
• Image and video generators in that payload (grok-imagine-* and similar) stay out of the chat picker. If the request fails, the pinned catalog remains the fallback, so Copilot and the other static vendors are unchanged.
• An id that models.dev does not know yet inherits context and thinking levels from the newest pinned sibling (grok-4.6, then 4.5, then 4.3) instead of the 128k generic shape.

Test plan

• node --test test/vendor-oauth-login.test.mjs test/anthropic-oauth-retry.test.mjs (38 passing), including a login whose /models body contains grok-4.7 and grok-imagine-image
• Sign in to xAI in Settings → Model configuration, refresh models, and confirm grok-4.7 is selectable and a turn no longer reports Vendor account does not offer model "grok-4.7"

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[vastsa]

Maintainer-side handling so this could be gated and landed:

• The head's workflow runs were sitting in action_required (first-time contributor), so nothing had been validated; I approved them.
• I merged the current main into the branch to satisfy Head contains latest base (head f3ece7fb), no force-push and no content changes.

Independent check of the premise: pi-ai 0.86.1's pinned xAI catalog ships only grok-4.3 / grok-4.5 / grok-4.6, so on main a signed-in xAI account really cannot select grok-4.7 — the change addresses a real gap. All four gates are green on the updated head.