gemma-4-e2b-it Online GPU deployment kustomization missing -it

This misnames downstream resources. Example for GPU [here](https://github.com/GoogleCloudPlatform/accelerated-platforms/blob/e9b6942408667f75f4a7ef780b0ae9dfcb13396a/platforms/gke/base/use-cases/inference-ref-arch/kubernetes-manifests/online-inference-gpu/vllm/l4-gemma-4-e2b-it/kustomization.yaml).

Error on deployment:

```
File "", line 198, in _run_module_as_main
File "", line 88, in _run_code
File "/usr/local/lib/python3.12/dist-packages/vllm/entrypoints/openai/api_server.py", line 2094, in
validate_parsed_serve_args(args)
File "/usr/local/lib/python3.12/dist-packages/vllm/entrypoints/openai/cli_args.py", line 289, in validate_parsed_serve_args
validate_chat_template(args.chat_template)
File "/usr/local/lib/python3.12/dist-packages/vllm/entrypoints/chat_utils.py", line 1012, in validate_chat_template
raise ValueError(
ValueError: The supplied chat template string (/gcs/google/gemma-4-e2b/chat_template.jinja) appears path-like, but doesn't exist!
```

Compare the affected online GPU kustomization with platforms/gke/base/use-cases/inference-ref-arch/kubernetes-manifests/online-inference-gpu/vllm/l4-gemma-4-e2b-it/kustomization.yaml. Check the vLLM deployment configuration against the reported /gcs/google/gemma-4-e2b/chat_template.jinja path, and verify that downstream resources use the -it model name and deploy without the missing-template error.

Written by the indexing model from the issue text.