images not passed to llama.cpp

When using a vision model hosted on llama.cpp (Qwen3.8-Flash-Next) I can upload an image via the llama.cpp webui but goose's read_image tool apparently fails (model reports "This tool result included an image that was omitted as the model does not support vision").

Look at the read_image tool implementation and how it passes images to the model. Compare with how llama.cpp's webui handles image uploads. Check the model's vision support and the data format expected. The fix likely involves ensuring the image data is correctly included in the tool result payload sent to the model.

Do mô hình lập chỉ mục viết ra từ nội dung của issue.