feat: Add optional path var for instance level bucket path

Terraform will perform the following actions:

  # module.wandb.data.google_compute_forwarding_rules.all will be read during apply
  # (depends on a resource or a module with changes pending)
 <= data "google_compute_forwarding_rules" "all" {
      + id    = (known after apply)
      + rules = (known after apply)
    }

  # module.wandb.module.wandb.helm_release.wandb will be updated in-place
  ~ resource "helm_release" "wandb" {
        id                         = "wandb-cr"
      ~ metadata                   = [
          - {
...
              - values         = jsonencode(
                    {
                      - name = "wandb"
                      - spec = <<-EOT
...
                                "bucket":
                                  "name": "andrew-levin-gcp-intense-goblin"
                                  "path": ""
                                  "provider": "gcs"
                                "cloudProvider": "gcp"
 ...

        # (1 unchanged block hidden)
    }

Plan: 0 to add, 1 to change, 0 to destroy.

The new helm release spec wasn't shown when running terraform apply, but the updates are propagated to the deployment as expected:

when bucket_path was NOT specified, when I exec'd to the wandb-app pod, I saw

wandb@wandb-app-65d8cb96-m7jb5:~$ env | grep -i bucket
BUCKET=gs://andrew-levin-gcp-intense-goblin

when bucket_path was specified as `"nested/path/test", I saw

wandb@wandb-app-7cbb475896-79266:~$ env | grep -i nested
BUCKET=gs://andrew-levin-gcp-intense-goblin/nested/path/test
"store": "gs://andrew-levin-gcp-intense-goblin/nested/path/test",
OVERFLOW_BUCKET_ADDR=gs://andrew-levin-gcp-intense-goblin/nested/path/test
GORILLA_FILE_STORE=gs://andrew-levin-gcp-intense-goblin/nested/path/test

wandb / terraform-google-wandb

feat: Add optional path var for instance level bucket path #145