Emitting diagnostic event containing Soroban resource utilization metrics

[jayz22]

Description

Resolves ##3759
This PR adds a few metrics to inform on network utilization and help answer the question "how close we are to the limits".
It adds the the metrics as diagnostic events, so that they can be consumed by the downstream.
It also adds a new "invoke time" metric which is a better comparison to the cpu instructions than the current "exec time", resolving #3776 (comment)

Checklist

• Reviewed the contributing document
• Rebased on top of master (no merge commits)
• Ran clang-format v8.0.0 (via make format or the Visual Studio extension)
• Compiles
• Ran all tests
• If change impacts performance, include supporting evidence per the performance document

[jayz22]

After discussing with @sisuresh, we decided it is better to just have one event per metric, with the topics containing {"core_metrics", "metric_name"}, more specifying info maybe included in the topics in the future. The downside is each tx/op will now emit 20 (may grow in the future too), instead of 1 extra events, and downstream will decide how to index them. This should be fine and it's better than us defining a fixed data structure (of a single large event) for the downstream to parse.
cc @tsachiherman

[graydon]

r+ f0f9337

[graydon]

LGTM