Fix deprecated kubernetes_namespace resource

[dhavaln-able]

Summary

Terraform's kubernetes provider deprecated the kubernetes_namespace resource in favor of kubernetes_namespace_v1 (part of the v2.0 provider naming convention change, where several resources gained _v1 suffixed replacements). Consumers of this module currently see:

Warning: Deprecated Resource

  with module.efs-csi-driver.kubernetes_namespace.kubernetes_efs_csi_driver,
  on namespace.tf line 1, in resource "kubernetes_namespace" "kubernetes_efs_csi_driver":
   1: resource "kubernetes_namespace" "kubernetes_efs_csi_driver" {

Deprecated; use kubernetes_namespace_v1.

Changes

• namespace.tf: kubernetes_namespace → kubernetes_namespace_v1
• helm.tf: update the matching depends_on reference
• versions.tf: bump the declared kubernetes provider constraint from >= 1.10.0 to >= 2.0, since kubernetes_namespace_v1 doesn't exist in provider versions before 2.0 — the old constraint would let someone pin an incompatible provider version and hit a hard error instead of just a warning
• namespace.tf: added a moved block (kubernetes_namespace.kubernetes_efs_csi_driver → kubernetes_namespace_v1.kubernetes_efs_csi_driver). Without it, Terraform treats the resource-type change as a brand new resource and plans to destroy the existing namespace (and everything running in it) and recreate it on upgrade — the moved block makes this a transparent state migration instead, so existing consumers see no diff on the namespace itself.
• versions.tf: bump required_version from >= 0.13 to >= 1.1, since moved blocks require Terraform 1.1+

Test plan

• terraform plan against a cluster consuming this module no longer shows the deprecated-resource warning
• terraform plan on an existing deployment shows no changes to the namespace itself (the moved block should make this a no-op state migration, not a destroy/recreate)