Revert enable_prefix_caching change - causes vLLM init failure

The enable_prefix_caching=False parameter causes vLLM to fail with:
'OSError: Can't load image processor for lightonai/LightOnOCR-2-1B'

Keeping only the max_tokens change (6144 -> 4096) to fix repetition.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

lighton-ocr2.py

@@ -340,7 +340,6 @@ def main(
         gpu_memory_utilization=gpu_memory_utilization,
         limit_mm_per_prompt={"image": 1},  # One image per prompt
         enforce_eager=False,  # Use torch.compile for better performance
-        enable_prefix_caching=False,  # Recommended by model card
     )
 
     # LightOnOCR-2 recommended sampling parameters