Ensure that data import spot fleet instances are unable to run the API http ECS tasks

[CloudNiner]

Unintentional downtime occurred when all of our ECS https tasks got placed onto data import instances while the import instances were running. When those instances were all forcibly killed because the import jobs were done, the HTTPS API tasks serving the public API also got killed.

We could address this in two ways. We could maintain a separate ECS cluster for the data import tasks. Or, we could add task placement constraints on the HTTPS tasks so that they're only placed on the smaller instances that are a permanent part of the current ECS cluster.