Added startupProbe endpoint to minimize metrics gaps

[jkroepke]

Hi,

i would like to ask if its possible to and an endpoint which observes if all defined metric collectors are run once.

Such an endpoint could be used for an startupProbe. Together with RollingUpgrade, Prometheus would reference to the existing running pod until the metrics from the new pods are available. This would minimize metrics gaps on Promethes and Grafana.

Since costs queries are enabled, the initial run could takes a while. The endpoint should take care of the cache, e.g. the costs cached an available, the endpoint should not wait for an initial cost query run.

[mblaschke]

will have a look how to implement this feature

[mblaschke]

i've added caching to all collectors, so this should speed up the the restart of the exporter
added with 23.5.0-beta0

[mblaschke]

is caching solving the issue?