Update RPC release eval endpoint load test configuration with insights from Overcat's data

[cjonas9]

Prerequisites: #860, #924

What

We now have data and analysis of expected RPC traffic and the distributions/patterns followed by various endpoints along many dimensions: RPS, query patterns, etc. This issue provides for integrating the insights into both stellar-rpc and stellar-rpc-blaster. Though this should be driven by the conclusions of #924, preliminarily, this may look like:

• for stellar-rpc-blaster:
  • align the request bodies of parameter-backed endpoints crafted for use in endpoint load tests with what's seen in the data
  • create a benchmark endpoint load-test profile that follows the RPS distributions seen by the data
    • since this profile will likely not change, but must align with what's truly achievable under the minimum hardware requirements, there should probably be an optional float scale parameter one can specify in the profile. One way to accomplish this could be to normalize the profile such that the sum of all of the endpoints RPS values equals 100, and a benchmark load level L is established as some scale=L. Then, in the load test, each endpoint i is sent at a maximum of RPS of RPS_i * L.
• for stellar-rpc:
  • integrate the new endpoint load test configuration changes into the endpoint load test
  • test robustly to establish reliable and consistent benchmarking expectations given the configurations