Handle grayscale images in expand2square

[isaac-vidas]

Fix error when handling grayscale images in expand2square:

Exception in TokenizerManager:
Traceback (most recent call last):
  File "/home/gcpuser/sglang/python/sglang/srt/managers/tokenizer_manager.py", line 61, in get_pixel_values
    image = expand2square(
  File "/home/gcpuser/sglang/python/sglang/srt/mm_utils.py", line 167, in expand2square
    result = Image.new(pil_img.mode, (width, width), background_color)
  File "/opt/conda/envs/sglang_flashinfer/lib/python3.9/site-packages/PIL/Image.py", line 2941, in new
    return im._new(core.fill(mode, size, color))
TypeError: color must be int or single-element tuple

[haotian-liu]

Thanks. It makes sense not to assume that they are in RGB.

[haotian-liu]

Maybe just move this if pil_img.mode to the outer block, in case there's something else broken when width==height. Anyway we will need RGB.

[isaac-vidas]

Makes sense!
I moved the conversion to her as we only need to convert it if the width and height are not the same + changed elif to if condition to support the change.

[haotian-liu]

Thanks.