Benchmark serve over Standard and Premium blob. #1447

We want to benchmark the mean and tail latencies associated with serving various index sizes (and vector dimensions) when reading posting lists directly from blob. We need benchmarks for standard and premium blob tiers.

This will give folks in the low-throughput, cold query scenario to estimate the query latencies.