Optimize config for safetensors loading

Remove redundant pytorch_model.bin, use safetensors only

FIX: Add proper test_fixed.py with from_pretrained support

FIX: Add proper README.md with from_pretrained support

FIX: Add proper modeling_pixeltext.py with from_pretrained support

FIX: Add proper config.json with from_pretrained support

Fix README.md for proper loading

Fix load_model.py for proper loading

Fix modeling_pixeltext.py for proper loading

Fix config.json for proper loading

Upload custom PaliGemma OCR model

initial commit