Reduce down time when performing avalon or avalon-docker upgrades

[cjcolvar]

Description

It seems like terraform will destroy the ec2 instance when changing either of the avalon or avalon-docker branch/tag/commit variables. This could leave the system down for 10-15 minutes while a new ec2 instance is created, the codebuild build to run and deploy, and finally for the docker containers to boot.

Done Looks Like

• Explore using terraform lifecycle properties like create_before_destroy to reduce the downtime when upgrading.