docs: improve monitoring documentation

[hiddeco]

Fixes #381.
Fixes #62.
Documents #383.

[stefanprodan]

I think we should include the list:

	// Unknown is mapped to 0
	v1.HelmReleasePhaseChartFetchFailed: -4,
	v1.HelmReleasePhaseFailed:           -3,
	v1.HelmReleasePhaseRollbackFailed:   -2,
	v1.HelmReleasePhaseRolledBack:       -1,
	v1.HelmReleasePhaseRollingBack:      1,
	v1.HelmReleasePhaseInstalling:       2,
	v1.HelmReleasePhaseUpgrading:        3,
	v1.HelmReleasePhaseChartFetched:     4,
	v1.HelmReleasePhaseSucceeded:        5,

[stefanprodan]

We could also include an example of how to alert on failed releases, similar to flux docs: https://docs.fluxcd.io/en/1.19.0/references/monitoring/

[hiddeco]

@stefanprodan yes, that was also my initial though but I was having a hard time finding the right formatting for all those values. Given the document is simple and table based at the moment.

[onedr0p]

@hiddeco You can include See phases below link and list them outside the table

[sa-spag]

If you want to mention Prometheus alert rules, I suggest the following:

• the Helm Operator struggles to process its release queue in a timely manner:

alert: HelmOperatorLowThroughput
expr: flux_helm_operator_release_queue_length_count > 0
for: 30m

• a HelmRelease was automatically rolled back:

alert: HelmReleaseRolledBack
expr: flux_helm_operator_release_phase_info == -1

• a HelmRelease is subject to an error:

alert: HelmReleaseError
expr: flux_helm_operator_release_phase_info < -1

Note that these does not cover Helm releases statuses as exposed by sstarcher/helm-exporter (you might want to include that notice in the docs).

[stefanprodan]

LGTM 💯