v0.1.0 — first release

First public release of dubfit.

Fits dubbed narration to a video scene by scene, capping the speed change and carrying the residual drift forward, so timing never slides and no scene is sounded rushed.

• Per-scene atempo fitting with drift carry-forward (no constant-ratio stretch, no accumulated desync)
• Final scene absorbs the remainder so the output lands exactly on the target length
• Reads scene timings from JSON (three shapes) or SRT cue boundaries
• dubfit plan to inspect the fit without rendering; dubfit fit to render (wav + optional m4a)
• No Python dependencies; ffmpeg on PATH required
• 29 tests, verified against a real 53-scene, 44-minute production run