Docs answer "what will this cost me" with a real number

Eric's clean run put a price on the pipeline's only paid step:
65 shelf photos, 136 games, under a dollar on the default model. The
README's vision requirement and the guide's config section both now
say so, plus the shape of the cost — one-time, since per-photo reads
are cached and re-runs are free — and that the BGG side costs nothing.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016jXZFSTZQKzAC8fqpWSz9g

docs/guide.md

@@ -32,6 +32,8 @@ Your credentials never leave your machine except to sign in to boardgamegeek.com
 
 Non-secret knobs live in `config.toml`: `photos_dir`, `data_dir`, the BGG rate limit, and the vision setup — a `[vision.<provider>]` block per provider ("anthropic" or any OpenAI-compatible endpoint, including a local [Ollama](https://ollama.com/)), with `vision_provider` picking one. Local models read spines noticeably worse than frontier ones — expect a longer proofread pass on the Titles page, not a broken pipeline.
 
+**What extraction costs:** the vision call is the pipeline's only paid step, and it's small — the author's full collection (65 shelf photos, 136 games) came to under a dollar on the default model (Claude Sonnet). It's also one-time: raw reads are cached per photo, so re-running extract is free, and only new or replaced photos are ever sent again. Everything BGG-side is free (the API token costs nothing).
+
 ## The web app
 
 ```sh