[Optimization] Use ECR's Pull Through Cache instead of pulling the image url directly

[Cameronsplaze]

Is your feature request related to a problem? Please describe.
This might speed up how long everything takes to turn on. Before doing this, we need tests and/or scripts to time how long it actually takes to get the service online. (If there's a tiny improvement for a LOT of complexity, it's not worth it).

Describe the solution you'd like
The idea is if we use Pull Through Cache, you'll always have the latest/greatest image available to EC2, and since ECR is in network, it should download it fast (assuming it already has the latest image)

I think this is something that should be optional, since

1. it adds cost since the image is stored in ECR, and you're paying for space.
2. You have to provide creds to be able to download from specific areas like GitHub/Docker. (That's what the docs say at least. Try it without, maybe it's just like pulling from them with the cli.)

Describe alternatives you've considered
This is an optional feature, the alternative is to just not do it and download directly from the source.

Additional context

• cdk construct
• Blog post here

[Cameronsplaze]

This one is blocked on the test-suite / timing-scripts for now.