Crate workshop: Running Evaluations

[deanwampler]

As part of the TSEI promotion plan, this workshop expands on the brief coverage of how to run benchmarks in #43.

Using a canned set of benchmarks, it will focus on the unified reference stack for running evaluations (e.g., a set of benchmarks) both offline during development and online during inference. What is different about these deployments, such as performance considerations and the kinds of evaluations that make sense in each context?