[REA] Multi-GPU Training Benchmarking

[bschifferer]

What questions are you trying to answer? Please describe. We have NVTabular + NVTabular dataloaders to train on multi-GPUs. We provide multiple examples how to scale to multi-GPU. We want to benchmark, how throughput scales with 1x, 2x, 4x, and 8x GPUs and profile any bottlenecks