Add basic telemetry to the cli tool

**Late night idea warning**

We need data to drive our decisions, and understanding of how people use pipelines.
To allow us to make data-driven decisions, let's add some basic telemetry to each of our commands.

What we want to know is:
1. Shape+form of pipelines? (pipeline JSON for Spinnaker)
2. What are the common failure errors?
3. Which backends are used?
4. How many saves have been done using the CLI?
5. What are tests run, and how do they fail?

This data **must** be anonymized.
We will be looking for patterns, not exact problems.

The design must allow removing the telemetry entirely from the compiled output for use-cases where telemetry is a problem.

No files, tests, or entry points are named. Start by locating the CLI command entry points and the code handling pipeline JSON, backends, saves, and tests, then determine how anonymization and compile-time removal could fit the existing build. Done means the telemetry scope is agreed, anonymization is verified, and builds can exclude it entirely.

Written by the indexing model from the issue text.