Finalizing Thesis Multimodal Architecture for Hugging Face

feat: implement fairseq compatibility shim and working audio inference pipeline

feat: initial project setup with audio/video detection modules and fusion