### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

## Actual behaviour

When sending `service_tier` (e.g. `"flex"`, `"priority"`) through the LiteLLM proxy to a `vertex_ai/` model, Vertex AI rejects the request with a 400 error:

Invalid value at 'service_tier' (type.googleapis.com/google.cloud.aiplatform.v1.ServiceTier), "flex"

This happens for ALL values — lowercase (`flex`, `priority`, `standard`), uppercase (`FLEX`, `PRIORITY`, `STANDARD`), and across all Gemini models tested (`gemini-3.5-flash-lite`, `gemini-2.5-flash`, `gemini-2.5-pro`, `gemini-3.5-flash`).

## Expected behaviour

`service_tier` should be correctly translated and accepted by the Vertex AI API, or gracefully dropped if the field is not supported for the v1 API / project configuration.

## Impact

Users cannot use `service_tier` to select priority/flex capacity for Vertex AI Gemini models. This forces workarounds like defining duplicate model entries in the proxy config with `extra_headers` (e.g. `X-Vertex-AI-LLM-Shared-Request-Type: priority`), which defeats the purpose of the `service_tier` parameter added in #24898.

…