Looking into MMSeqs MSA input into AF3 pipeline code from AlphaFast

[keiran-rowell-unsw]

AlphaFast: ultra-high-throughput AlphaFold3 inference with MMSeqs2-GPU: https://github.com/RomeroLab/alphafast/tree/main

[keiran-rowell-unsw]

NVIDIA NIM also did this. Did cursory look but not worth implementing in Hackathon since the focus is on Boltz