[Search Sessions] Error state needs retry mechanism

[Dosant]

In case of search status request fail due to network or cluster instability, we assume ERROR_STATE as the same state when es returned search error:

kibana/x-pack/plugins/data_enhanced/server/search/session/get_search_status.ts

Lines 46 to 55 in 821aeb1

	return {
	status: SearchStatus.ERROR,
	error: i18n.translate('xpack.data.search.statusThrow', {
	defaultMessage: `Search status threw an error {message} ({errorCode}) status`,
	values: {
	message: e.message,
	errorCode: e.statusCode || 500,
	},
	}),
	};

In this case, we transition the search session to an error state and then stop checking its search statuses. But it actually might be that the searches will successfully finish later. Also, searches might leave hanging, executing and loading the cluster: #105434

We need to handle that catch close differently. Maybe introduce a retry mechanism, where we would try to check that search status again on the next task run.

[elasticmachine]

Pinging @elastic/kibana-app-services (Team:AppServices)

[elasticmachine]

Pinging @elastic/kibana-data-discovery (Team:DataDiscovery)