Introduce LanguageModelToolUse::raw_input (#29322)

This is to enable alternative streaming solutions at the application
layer. I'm not sure we really should have performed parsing of the input
at this layer. Either way I want to experiment with streaming approaches
in a separate crate on a branch, and this will help.

/cc @maxdeviant @bennetbo @rtfeldman

Closes #ISSUE

Release Notes:

- N/A

crates/google_ai/src/google_ai.rs

@@ -338,6 +338,7 @@ pub struct CountTokensResponse {
 #[derive(Debug, Serialize, Deserialize)]
 pub struct FunctionCall {
     pub name: String,
+    pub raw_args: String,
     pub args: serde_json::Value,
 }