Investigate telemetry improvements around how users use workloads

Right now we have very little visibility into how users use workloads. With the advent of workload sets in 8.0.400 and beyond, we want to collect data on what modes users choose and what versions of workloads they are using. Potential data points:

## Feature set usage
* loose manifest vs workload-set usage rates
* workload set usage kind - explicitly pinned in global.json, implicitly pinned by the SDK, explicitly pinned by CLI command
* rates of pinning to workload set versions that are behind/ahead of the current SDK. essentially asking the question "how often do people bump SDKs but lag on workloads and why?"
* logging versions of workloads installed and/or used by projects

## Operational analysis/performance
* success rates for installing/restoring/updating workloads
* reasons for failures - misconfigured feeds? missing versions? anything else?