agent: Preserve thinking blocks between requests (#29055)

Looks like the required backend component of this was deployed.

https://github.com/zed-industries/monorepo/actions/runs/14541199197

Release Notes:

- N/A

---------

Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Agus Zubiaga <hi@aguz.me>
Co-authored-by: Richard Feldman <oss@rtfeldman.com>
Co-authored-by: Nathan Sobo <nathan@zed.dev>

crates/language_model/src/language_model.rs

@@ -65,9 +65,14 @@ pub struct LanguageModelCacheConfiguration {
 pub enum LanguageModelCompletionEvent {
     Stop(StopReason),
     Text(String),
-    Thinking(String),
+    Thinking {
+        text: String,
+        signature: Option<String>,
+    },
     ToolUse(LanguageModelToolUse),
-    StartMessage { message_id: String },
+    StartMessage {
+        message_id: String,
+    },
     UsageUpdate(TokenUsage),
 }
 
@@ -302,7 +307,7 @@ pub trait LanguageModel: Send + Sync {
                             match result {
                                 Ok(LanguageModelCompletionEvent::StartMessage { .. }) => None,
                                 Ok(LanguageModelCompletionEvent::Text(text)) => Some(Ok(text)),
-                                Ok(LanguageModelCompletionEvent::Thinking(_)) => None,
+                                Ok(LanguageModelCompletionEvent::Thinking { .. }) => None,
                                 Ok(LanguageModelCompletionEvent::Stop(_)) => None,
                                 Ok(LanguageModelCompletionEvent::ToolUse(_)) => None,
                                 Ok(LanguageModelCompletionEvent::UsageUpdate(token_usage)) => {