Consider adding telemetry to the playground

It would be really nice to have aggregations of error data from both the Azure and core playgrounds that we could use to measure usage and find bugs. We would want to make sure we have a directed purpose for the data, a definite set of questions we wanted the answer to. Some possibilities:
- Catching compiler crashes
- Playground fails to even load
- Finding frequency of different kinds of errors, looking to shape tooling and improve error messages

Start by reviewing the Azure and core playgrounds and the proposed telemetry questions: compiler crashes, load failures, and error frequencies. Define the directed purpose and definite questions before deciding what data should be collected. Done means an agreed telemetry design with a clear purpose and scope.

Written by the indexing model from the issue text.