use kustomize instead of sed for environment configuration

[hahn-kev]

Fixes #1783

Description

refactored our k8s resources to use kustomize to simplify having different configs per environment

Checklist

• I have labeled my PR with: bug, feature, engineering, security fix or testing
• I have performed a self-review of my own code
• I have reviewed the title & description of this PR which I will use as the squashed PR commit message
• I have commented my code, particularly in hard-to-understand areas
• I have added tests that prove my fix is effective or that my feature works
• I have enabled auto-merge (optional)

to test this I used kubectl --context dallas-rke diff -k . in the staging folder (and similar for prod) to verify that nothing unexpected changes.

[github-actions]

Unit Test Results

362 tests   362 ✔️  12s ⏱️
  37 suites      0 💤
    1 files        0 ❌

Results for commit 05c4242.

[rmunn]

Still reading through this, but one comment I have is that you have commands like kubectl --context aws-rke in the Makefile. That's a good idea, but we'll need to add something to the README saying "You need to name your kubectl contexts by these names" or else people who have a different name in their .kubectl will have the command fail by default. And I'd also suggest changing the context names to something that includes "languageforge" or "lf" in it, because in my .kubectl I have the following:

- context:
    cluster: aws-rke
    namespace: languagedepot
    user: aws-rke
  name: languagedepot-rke
- context:
    cluster: aws-rke
    namespace: languageforge
    user: aws-rke
  name: languageforge-rke

I should actually rename those to something like language{forge,depot}-prod or {lf,ld}-prod but haven't gotten around to it. The point is that the kubectl context can carry a default namespace along with it, and I use that so I don't see languageforge pods when I'm deploying languagedepot, and vice versa.

[rmunn]

Looks good to me. I believe the GHA workflows will continue to work with no modifications with this approach, though of course we'll need to see it in action to be 100% certain. But it looks like you have everything covered that I could think of.

[hahn-kev]

That's a good point about the context names. I'll update the readme.