Improve scope and consistency of Prometheus exports

## What?
Many of the metrics gathered in GL are not exported via Prometheus. It's unclear how these were selected for inclusion or omission; there are definitely useful metrics that are currently inaccessible.

- a specific metric should be available across all input types (if applicable)
- any metric that is exposed on `get /cluster/{nodeId}/metrics/names` should also be exported to Prometheus
- exported metrics should be consistent across core, enterprise and cloud

## Why?
It looks like a bug, when a useful and generic metric like `gl_input_written_bytes_one_sec` is only available for GELF HTTP, but not for GELF UDP or any other input type (#11108).

Users should be able to easily access and monitor all available metrics. It should not require filing an issue and making a code change to expose a specific metric.