MetadataIndexAliasesService does not respect timeout specified in request when submitting index-aliases tasks

[PeteGillinElastic]

In MetadataIndexAliasesService we submit tasks to a queue to apply the metadata changes to create/delete aliases. There is a timeout in the request (IndicesAliasesClusterStateUpdateRequest.masterNodeTimeout()) but it is not used here, the task has a null i.e. infinite timeout. As a consequence, if this originated from a REST request, the user will eventually see an error when something times out in the serving infrastructure, but this task can stay pending indefinitely. (We have seen a case where a bug meant that those tasks were not getting processed, and we had a large number of tasks which had been pending for several days.) There is a TODO in the code for this.

[elasticsearchmachine]

Pinging @elastic/es-data-management (Team:Data Management)