Dual cluster FRIDGE deployment

[craddm]

This PR splits FRIDGE deployment up across two clusters.

The access cluster contains an SSH server and the Harbor container registry.

The isolated cluster contains the rest of the FRIDGE deployment.

This complements PR #131, which does the initial deployment of two AKS clusters.

The intended deployment pattern here is for the TRE Operator to first deploy the access cluster components, then set up the SSH proxy to allow them to deploy the isolated cluster components.

That will then be followed by a final infrastructure configuration step to limit network traffic more completely.

[JimMadge]

How much of this is copied from the existing Pulumi project?

Wouldn't it make more sense to replace that?

[craddm]

It will do, once I'm happy with it :)

[craddm]

Although I do wonder if it's helpful to keep a single-cluster spec around for some of the local dev work - a lot easier to run a single cluster on a VM locally than to run two...

[JimMadge]

Although I do wonder if it's helpful to keep a single-cluster spec around for some of the local dev work - a lot easier to run a single cluster on a VM locally than to run two...

I think that is running just the isolated cluster part in a K3s instance. For local dev you don't need the access cluster or lock down.

That way, at least that part is the same thing locally and when deployed.

[JimMadge]

It will do, once I'm happy with it :)

Someone needs to tell you about this version control thing. You don't need to make a copy of all your code when you want to make changes any more 😜