Changelog

Fix a bug where training calendar invites were being sent for 7am MT
Fix an edge case bug where a patient appointment was created twice in Stat, when a customer sent multiple HL7 messages to the same appointment within a second
Clean up to some scaffolding created for the conversion project (no effect to any logic in use)
Tweak to the scaling of our infrastructure so it is more likely to scale up to handle increased loads

Updates to the production infrastructure configurations, which will keep things more scaled up while we evaluate the overall size needed of the infrastructure
Removing automatic restarts if a device has not checked in within the last 10 minutes
Fixing a bug where if we received two appointment updates for the same patient within seconds of each other, only one was processed, and the other was discarded (now they will run after each other, every 5 seconds)