Add metrics to usage of cache in handlers

[jamestelfer]

The handlers use a cache with the PipelineID as a key that will return the same token on repeated requests for a given pipeline.

We need to change our telemetry such that:

• the current span has a hit/miss indicator for cache usage
• there is a special case around repository URL handling that causes a cache miss: annotate this on the span also
• add cache metrics so we can track hit rate over time. Pipeline name or ID would be useful as attributes for this metric.