Upgrade rfdetr to v1.6.5.post2 and migrate to proper segmentation checkpoints

Replace git dependency with pinned PyPI package for rfdetr

fix colors

add rf-detr seg

bump to 1.3.0

Update app.py

set medium as the default checkpoint

Update app.py to allow new nano, small, and medium checkpoints

use RF-DETR 1.2.0 release code

fix: 🐞 gradio fix for space to work. (#2)

add progress; drop example cache

drop progress

initial video processing support

add video files

test video processing on HF spaces

update UI to support video inference

drop `spaces` call; use GPU runner for now

allow users to pick inference resolution

checkpoint selection added

initial commit

initial commit