Remove vision OCR — use Tesseract-only pipeline for receipt parsing

The llama3.2-vision model was producing unreliable structured data
(wrong vendors, amounts, dates) making expense reports worse than
Tesseract + LLM extraction.  Removes _ocr_image_vision(), the
vision JSON fast path in _parse_receipt_text(), _match_category(),
and the vision_ocr_model config setting entirely.

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

agent_service/config.py

@@ -16,10 +16,6 @@ class Settings(BaseSettings):
     ollama_model: str = 'activeblue-chat'
     ollama_timeout: int = 300
     ollama_max_concurrent: int = 2
-    # Set to a vision-capable model (e.g. llama3.2-vision:11b) to use
-    # vision OCR for receipt images instead of Tesseract.  Leave empty
-    # to keep the Tesseract pipeline.
-    vision_ocr_model: str = ''
 
     # Anthropic / Claude
     anthropic_api_key: str = ''