ogar-vocab: declare the tesseract-rs OCR capability surface as the authoritative action table

tesseract-rs's pure-Rust OCR transcode has no source AST to lift ActionDefs
from (unlike every other producer in this workspace), so its eight
capabilities (recognize_line / recognize_page / extract_text_layer /
extract_page_image / render_text / render_tsv / render_hocr /
render_searchable_pdf) are declared directly in a new ocr_actions module —
real ActionDef values (identity/predicate/object_class/kausal) targeting
the already-minted 0x08XX class_ids concepts (textline/page_image/
ocr_renderer), plus a dependency-free OcrActionParam mirror of
ogar-from-schema's arago-parity ActionParam (kept local to avoid a cyclic
dep, since ogar-from-schema already depends on ogar-vocab).

No codebook/class_ids entries were added or changed — this table
consumes the existing 0x0805/0x0808/0x0809 mints from PR #172.

Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016b33swuXE23hKtqxsHu9p1

Author: claude

crates/ogar-vocab/src/lib.rs

@@ -34,6 +34,15 @@ use serde::{Deserialize, Serialize};
 /// `<port>::<path>(<shape>)` grammar (`E-GRAMMAR-IS-THE-RECIPE-SHAPE`).
 pub mod recipe;
 
+/// The tesseract-rs OCR capability surface — a hand-authored, non-`lift_*`
+/// [`ActionDef`] table (tesseract-rs has no source AST to extract from; see
+/// the module doc for why this is the authoritative action table rather
+/// than an extraction). Declares `recognize_line` / `recognize_page` /
+/// `extract_text_layer` / `extract_page_image` / `render_text` /
+/// `render_tsv` / `render_hocr` / `render_searchable_pdf`, each targeting a
+/// minted `0x08XX` [`class_ids`] concept.
+pub mod ocr_actions;
+
 /// Source language hint — discriminates the producer for traceability
 /// and for emitter dispatch on Ruby/Python-specific extension shapes
 /// (e.g. Odoo `ComputedField`). Not a hard schema discriminator: a class