Add Prometheus ServiceMonitor discovery as an input

[getsomebread]

Currently, when scraping Prometheus metrics, need to manually define all endpoints to scrape, which can vary between clusters. Instead, would be nice to tell Fluent Bit to scrape all discovered ServiceMonitors so one Fluent Bit config would work for all clusters and new metrics get scraped as they are added.

This functionality exists in Alloy, see https://grafana.com/docs/alloy/latest/reference/components/prometheus/prometheus.operator.servicemonitors/