[ResponseOps][Task Manager] when changing plugin status, indicate the reason in a logged message

[pmuellr]

We've gotten comments on the task manager health report, and the messages task manager is logging, that it's often unclear why task manager might be "complaining" about something, and what the actual cause is.

As an example, when task manager status goes from green to yellow, there is often no reason given why it it did that. Looking through the task manager diagnostics also doesn't yield obvious clues.

At a minimum, it feels like whenever TM realizes it's going into some warning state, it should log a message indicating why. Today, I believe in many cases, we already do debug log a message - perhaps we could just promote these to warnings.

For example:

kibana/x-pack/plugins/task_manager/server/monitoring/capacity_estimation.ts

Lines 234 to 255 in 340d592

	function getHealthStatus(logger: Logger, params: GetHealthStatusParams): HealthStatus {
	const {
	assumedRequiredThroughputPerMinutePerKibana,
	assumedAverageRecurringRequiredThroughputPerMinutePerKibana,
	capacityPerMinutePerKibana,
	} = params;
	if (assumedRequiredThroughputPerMinutePerKibana < capacityPerMinutePerKibana) {
	return HealthStatus.OK;
	}
	if (assumedAverageRecurringRequiredThroughputPerMinutePerKibana < capacityPerMinutePerKibana) {
	logger.debug(
	`setting HealthStatus.Warning because assumedAverageRecurringRequiredThroughputPerMinutePerKibana (${assumedAverageRecurringRequiredThroughputPerMinutePerKibana}) < capacityPerMinutePerKibana (${capacityPerMinutePerKibana})`
	);
	return HealthStatus.Warning;
	}
	logger.debug(
	`setting HealthStatus.Error because assumedRequiredThroughputPerMinutePerKibana (${assumedRequiredThroughputPerMinutePerKibana}) >= capacityPerMinutePerKibana (${capacityPerMinutePerKibana}) AND assumedAverageRecurringRequiredThroughputPerMinutePerKibana (${assumedAverageRecurringRequiredThroughputPerMinutePerKibana}) >= capacityPerMinutePerKibana (${capacityPerMinutePerKibana})`
	);
	return HealthStatus.Error;
	}

[elasticmachine]

Pinging @elastic/response-ops (Team:ResponseOps)

[pmuellr]

Was noted in triage, that we think the Kibana health APIs allow setting a "reason" or similar as well, so we should do that as well, if possible.