## Summary

Related to elastic#144161

Found that on a bulk update tags task failure, the task didn't stop
after 3 retries (should be over in less then a minute), the retries kept
happening for 2 hours.
This change removes the retry task if 3 retries are reached.

Also testing in cloud deployment to see if the tags error can be
reproduced with this fix.
I could reproduce the reported error locally, and seeing it goes away
with this fix.

To verify:
- Add at least 50k agents with the `create_agents` script in kibana repo
- open Kibana, select the 50k agents, and open Actions / Add tags
- Try this in a few seconds: add 2 new tags, and remove one of them
- Wait about 30s, the agents should reflect the changes
- Check the logs to see that the tasks are removed after 3rd retry is
reached or successful.
- Check that there are no more running tasks. Any running task can be
found in Kibana Console by running this query: `GET
.kibana_task_manager/_search?q=task.taskType:"fleet:update_agent_tags:retry"`

Locally simulated an error to test that the retry (and check) task is
removed:

```
[2022-12-07T15:52:16.415+01:00][ERROR][plugins.fleet] Retry elastic#3 of task fleet:update_agent_tags:retry:848984ab-c11d-4ebe-8d1f-606143dd656b failed: failing task
[2022-12-07T15:52:16.416+01:00][WARN ][plugins.fleet] Stopping after 3rd retry. Error: failing task
[2022-12-07T15:52:16.416+01:00][INFO ][plugins.fleet] Removing task fleet:update_agent_tags:retry:check:848984ab-c11d-4ebe-8d1f-606143dd656b
[2022-12-07T15:52:16.416+01:00][INFO ][plugins.fleet] Removing task fleet:update_agent_tags:retry:848984ab-c11d-4ebe-8d1f-606143dd656b
```

(cherry picked from commit 431c32b)