Use SGMV for prefill BGMV for decode

[tgaddair]

Closes #333.

There were broadly two main issues affecting LoRAX throughput for single-adapter performance vs vLLM:

1. Use of CUDA graph compilation by default
2. Use of BGMV kernel during decode

In this PR, we change BGMV to be the default during decode, and apply it to CUDA graph mode. Additionally, we retain SGMV for prefill (non-CUDA graph) and have a just-in-time tracing of specific LoRA layers to avoid having to replay computation for unused LoRA layers. All in all, we are now a good bit ahead of vLLM performance on single LoRA inference, and continue to even better at multi-LoRA scale. We further find that using Medusa gives an additional boost to performance that makes LoRA inference faster than base model performance (no adapter).

vllm + compile (baseline): 61 tokens/s
lorax (baseline, sgmv only): 52 tokens/s
lorax + bgmv: 59 tokens/s
lorax + bgmv + compile: 65 tokens/s
lorax + bgmv + compile + medusa: 73 tokens/s