0.7.0

Added

• "Caption the whole video" mode. A second way to work alongside AI clip
  finding: give it a 16:9 video and it comes back as one vertical, captioned
  edit you can trim, restyle and export, with optional speaker tracking and
  auto zoom. Both modes work on the same project and share the transcript, so
  switching between them never pays for transcription twice.
• Full-quality WorkVivo uploads. Clips now upload through the presigned
  flow rather than inline through the Customer API, which rejected anything
  beyond a few megabytes. Needs a one-off browser sign-in in Settings; the API
  remains the fallback.
• Custom caption fonts. Upload any TTF or OTF. Families are matched on the
  name embedded in the file rather than the filename, so previews and exports
  agree.
• Size-targeted rendering. When an upload has a hard byte cap, the bitrate
  is planned up front and the clip is encoded once to hit it, instead of
  rendering at a quality target and re-compressing afterwards.

Fixed

• Burned-in captions rendered at roughly 58% of their intended size. CSS
  font-size sets the em square, but libass sizes text against the font's OS/2
  window ascent plus descent. Em sizes were being passed straight through as
  ASS Fontsize. This was the gap between the live preview and the exported
  file.
• Very long edits could build zoom filter graphs large enough to choke ffmpeg.
  Zoom events are now capped per clip.