assistant: Stream tool uses as structured data (#17322)

This PR adjusts the approach we use to encoding tool uses in the
completion response to use a structured format rather than simply
injecting it into the response stream as text.

In #17170 we would encode the tool uses as XML and insert them as text.
This would require then re-parsing the tool uses out of the buffer in
order to use them.

The approach taken in this PR is to make `stream_completion` return a
stream of `LanguageModelCompletionEvent`s. Each of these events can be
either text, or a tool use.

A new `stream_completion_text` method has been added to `LanguageModel`
for scenarios where we only care about textual content (currently,
everywhere that isn't the Assistant context editor).

Release Notes:

- N/A

crates/language_model/src/provider/ollama.rs

@@ -13,6 +13,7 @@ use std::{collections::BTreeMap, sync::Arc, time::Duration};
 use ui::{prelude::*, ButtonLike, Indicator};
 use util::ResultExt;
 
+use crate::LanguageModelCompletionEvent;
 use crate::{
     settings::AllLanguageModelSettings, LanguageModel, LanguageModelId, LanguageModelName,
     LanguageModelProvider, LanguageModelProviderId, LanguageModelProviderName,
@@ -302,7 +303,7 @@ impl LanguageModel for OllamaLanguageModel {
         &self,
         request: LanguageModelRequest,
         cx: &AsyncAppContext,
-    ) -> BoxFuture<'static, Result<BoxStream<'static, Result<String>>>> {
+    ) -> BoxFuture<'static, Result<BoxStream<'static, Result<LanguageModelCompletionEvent>>>> {
         let request = self.to_ollama_request(request);
 
         let http_client = self.http_client.clone();
@@ -335,7 +336,13 @@ impl LanguageModel for OllamaLanguageModel {
             Ok(stream)
         });
 
-        async move { Ok(future.await?.boxed()) }.boxed()
+        async move {
+            Ok(future
+                .await?
+                .map(|result| result.map(LanguageModelCompletionEvent::Text))
+                .boxed())
+        }
+        .boxed()
     }
 
     fn use_any_tool(