feat(ocr): auto-insert [unleserlich] markers for low-confidence words

New confidence.py module with two functions:
- apply_confidence_markers(): replaces words below threshold with
  [unleserlich], collapses adjacent markers into one
- words_from_characters(): reconstructs word-level confidence from
  Kraken's character-level data

Surya 0.17 provides native word-level confidence via line.words.
Kraken 7.0 provides per-character confidences via record.confidences.
Both engines now pass word+confidence data through main.py, which
applies the marker post-processing before returning the API response.

Threshold configurable via OCR_CONFIDENCE_THRESHOLD env var (default 0.3).
Frontend already renders [unleserlich] markers via transcriptionMarkers.ts.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

ocr-service/engines/kraken.py

@@ -37,6 +37,7 @@ def extract_blocks(images: list, language: str = "de") -> list[dict]:
     Coordinates are normalized to [0, 1].
     """
     from kraken import blla, rpred
+    from confidence import words_from_characters
 
     if _model is None:
         raise RuntimeError("Kraken model is not loaded")
@@ -73,6 +74,10 @@ def extract_blocks(images: list, language: str = "de") -> list[dict]:
             # Approximate polygon to quadrilateral
             quad = _approximate_to_quad(polygon_pts, page_w, page_h) if polygon_pts else None
 
+            # Extract word-level confidence for [unleserlich] marking
+            char_confidences = getattr(record, "confidences", [])
+            words = words_from_characters(record.prediction, char_confidences)
+
             all_blocks.append({
                 "pageNumber": page_idx,
                 "x": x1 / page_w,
@@ -81,6 +86,7 @@ def extract_blocks(images: list, language: str = "de") -> list[dict]:
                 "height": (y2 - y1) / page_h,
                 "polygon": quad,
                 "text": record.prediction,
+                "words": words,
             })
 
     return all_blocks

ocr-service/engines/surya.py

@@ -51,6 +51,17 @@ def extract_blocks(images: list, language: str = "de") -> list[dict]:
                     for p in line.polygon
                 ]
 
+            # Extract word-level confidence for [unleserlich] marking
+            words = []
+            if hasattr(line, "words") and line.words:
+                for word in line.words:
+                    words.append({
+                        "text": word.text,
+                        "confidence": word.confidence,
+                    })
+            else:
+                words = [{"text": line.text, "confidence": getattr(line, "confidence", 1.0)}]
+
             all_blocks.append({
                 "pageNumber": page_idx,
                 "x": x1 / page_w,
@@ -59,6 +70,7 @@ def extract_blocks(images: list, language: str = "de") -> list[dict]:
                 "height": (y2 - y1) / page_h,
                 "polygon": polygon,
                 "text": line.text,
+                "words": words,
             })
 
     return all_blocks