feat(ocr-pipeline): add LLM-based OCR correction step (Step 6)

Replace the placeholder "Koordinaten" step with an LLM review step that
sends vocab entries to qwen3:30b-a3b via Ollama for OCR error correction
(e.g. "8en" → "Ben"). Teachers can review, accept/reject individual
corrections in a diff table before applying them.

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

admin-lehrer/app/(admin)/ai/ocr-pipeline/page.tsx

@@ -8,7 +8,7 @@ import { StepDewarp } from '@/components/ocr-pipeline/StepDewarp'
 import { StepColumnDetection } from '@/components/ocr-pipeline/StepColumnDetection'
 import { StepRowDetection } from '@/components/ocr-pipeline/StepRowDetection'
 import { StepWordRecognition } from '@/components/ocr-pipeline/StepWordRecognition'
-import { StepCoordinates } from '@/components/ocr-pipeline/StepCoordinates'
+import { StepLlmReview } from '@/components/ocr-pipeline/StepLlmReview'
 import { StepReconstruction } from '@/components/ocr-pipeline/StepReconstruction'
 import { StepGroundTruth } from '@/components/ocr-pipeline/StepGroundTruth'
 import { PIPELINE_STEPS, type PipelineStep, type SessionListItem } from './types'
@@ -155,7 +155,7 @@ export default function OcrPipelinePage() {
     3: 'Spalten',
     4: 'Zeilen',
     5: 'Woerter',
-    6: 'Koordinaten',
+    6: 'LLM-Korrektur',
     7: 'Rekonstruktion',
     8: 'Validierung',
   }
@@ -173,7 +173,7 @@ export default function OcrPipelinePage() {
       case 4:
         return <StepWordRecognition sessionId={sessionId} onNext={handleNext} goToStep={goToStep} />
       case 5:
-        return <StepCoordinates />
+        return <StepLlmReview sessionId={sessionId} onNext={handleNext} />
       case 6:
         return <StepReconstruction />
       case 7: