check_drain potentially failing

[pjenvey]

check_drain has been seen to fail harmlessly (albeit rarely) on stage. We should boost its MaxAttempts from the current value of 10 to say 20.

[rpappalax]

see:
https://console.aws.amazon.com/states/home?region=us-east-1#/executions/details/arn:aws:states:us-east-1:142069644989:execution:ardere-dev-ardere:messages_partner_only___10_90___20_200___A7

[rpappalax]

the next one failed too...
https://console.aws.amazon.com/states/home?region=us-east-1#/executions/details/arn:aws:states:us-east-1:142069644989:execution:ardere-dev-ardere:messages_partner_only___10_90___60_600___A8

does this have to do with losing count of the expected instances?

[rpappalax]

{
  "error": "UndrainedInstancesException",
  "cause": "{\"stackTrace\": [[\"/var/task/handler.py\", 52, \"check_drain\", \"return AsynchronousPlanRunner(event, context).check_drained()\"], [\"/var/task/ardere/step_functions.py\", 368, \"check_drained\", \"raise UndrainedInstancesException(\\\"Services still draining\\\")\"]], \"errorType\": \"UndrainedInstancesException\", \"errorMessage\": \"Services still draining\"}"
}```