Consider optimizing stale telemetry handling in light of batching

[akhenry]

Is your feature request related to a problem? Please describe.
When processing a batch of telemetry, we probably only care whether the last parameter value in that batch is stale. If another parameter has come in after a stale parameter, then by definition it's no longer stale.

Optimization was identified during the implementation of telemetry batching, but was deemed to risky to implement in the initial round so is left as a future optimization.