`gemini-3-flash-preview` returns HTTP 404 on Vertex AI

Hi @jhk0530 , when using `token.vertex()` with `model_id = "3-flash-preview"`, the subsequent `gemini.vertex()` call fails with an HTTP 404 Not Found error. The same workflow works correctly for `2.5-flash` and `2.5-pro`. **To Reproduce** ```r library(gemini.R) # Works fine token <- token.vertex(jsonkey = "token/confidentials.json", model_id = "2.5-flash", region = "us-central1") gemini.vertex("hi", tokens = token) # Works fine token <- token.vertex(jsonkey = "token/confidentials.json", model_id = "2.5-pro", region = "us-central1") gemini.vertex("hi", tokens = token) # Fails with HTTP 404 token <- token.vertex(jsonkey = "token/confidentials.json", model_id = "3-flash-preview", region = "us-central1") gemini.vertex("hi", tokens = token) ``` **Error Message** ``` Error in `req_perform()`: ! HTTP 404 Not Found. Run `rlang::last_trace()` to see where the error occurred. ``` **Additional context** Authentication succeeds for `3-flash-preview` as well, which suggests the token generation step is fine.