Add a relation for input prometheus plugin

[Benjamin Allot]

Goal:
Telegraf has the ability to scrape prometheus metrics so it can expose them: https://github.com/influxdata/telegraf/tree/master/plugins/inputs/prometheus

This would allow a better homogeneous experience with metrics and labels, as telegraf would be the only point of scraping on a said node in the events that several process would expose metrics on a server.

Examples:
An example use case. If you need to scrape a list of prometheus urls you would create a file /etc/telegraf/telegraf.d/inputs_prometheus.conf like this:
```
[[inputs.prometheus]]
  urls = ["http://localhost:9145/metrics"]
```

However, one could be also interestd if scraping such metrics succeeded or failed.
Another file /etc/telegraf/telegraf.d/inputs_internal.conf could look like this:

```
[[inputs.internal]]
```
This would expose internal telegraf metrics such as the "internal_gather_errors" that, with the right label could be used to alert on scraping error.
i.e:
internal_gather_errors{alias="",host="prod:unit-0",input="prometheus",version="1.12.3"} 0

I therefore propose to add an interface layer to allow charm to publish input prometheus urls to the telegraf charm.
I would prefer to tie the internal to this one but this could also be separated.

[Eric Chen (eric-chen)]

Soon or later, the monitoring system will migrate to COS. Telegraf will be replaced by grafana-agent.
(https://charmhub.io/topics/canonical-observability-stack)
This feature won't be implemented.