Lectern v2.9.0

Added

• The home view shows the run's estimated token count next to the price.
  The ceiling most people meet first is a quota, not a bill.

Changed

• Rate limits are visible instead of silent: a throttled run used to
  look like a hung one. Every wait now appears in the activity log —
  "Gemini rate limit reached — waiting 30s (retry 1 of 5)" — and the
  server's own retry-after is honoured over Lectern's backoff.

Fixed

• When the retries run out, the error no longer promises a retry that is
  not coming. It tells apart a spent per-minute limit from an exhausted
  daily quota — waiting only helps for one of them — and says plainly that
  the cards generated so far are kept.