Implement rolling update for instances.

[finchr]

Describe the Feature

While updating the instance_type recently in preparation for a major version upgrade, both instances upgraded in parallel resulting in significant downtime. I found a simple fix for this which I will submit as a pull request.

Expected Behavior

At least one new node is in service at all times.

Use Case

A Zero or minimal downtime deploy.

Describe Ideal Solution

A rolling update.

Alternatives Considered

I considered a blue/green update which I was even able to implement using the create_before_destroy. I can provide this implementation if anyone is interested.

Additional Context

No response