SAM3 Slicer CLI Scripts

[BryonLewis]

resolves #337

background:

• initial model seems to be around 3.5GB, maybe tehy will have smaller model in the future
• Request a 1.4MB file bpe_simple_vocab_16e6.txt.gz

Notes:

• New API/system seems to require loading the entire video into memory and starting a session
  • May make sense to ffmpeg and clip out frames instead of loading entire video?
• Text Prompt is quite memory instensive and slow
• HuggingFace/Transformer models
  • Currently isn't in the pypi package, need to install from main source

TODO:

• Add frame ranges for script, so it can iterate over multiple frames
• Add support for either a text prompt or an existing bbox/mask
• Test script uploading to server and visualization
• Start working on updated SlicerCLI script for utilization of text prompt
• See if I can use one script with text prompt, existing box, or existing mask
• Attempt to see if the SlicerCLI script and interact with the tools for providing realtime updates of task running (I.E masks come in as they are complete)