Add Instrumentation to Runner

[darunrs]

While Grafana/Promethus provides general metrics to assess performance and faults, the granularity of data provided is not sufficient to understand performance characteristics of Runner. We will want to know in greater detail what is contributing to the latency of a runFunction execution, so that we can both detect and prioritize improvements to the workflow, aiming for the fastest backfill performance we can achieve.

The instrumentation can be gradually improved as we tackle various latency contributors.

Ideally, we will also integrate the instrumentation library into Grafana (and/or Prometheus) to reuse existing metrics infrastructure.

[darunrs]

We don't have any intentions to release this into prod. We may release it to dev briefly to understand latency impacts in the future, but local tracing provided the insights we needed for now, to improve latency.