WIP: Wiring context_new into agent — turn loop, StreamToken, dead code removal

Work in progress. New turn loop uses ResponseParser + StreamToken.
Killed StreamEvent, append_streaming, finalize_streaming, streaming_index,
assemble_api_messages, working_stack. Many methods still reference old
types — fixing next.

Co-Authored-By: Proof of Concept <poc@bcachefs.org>

src/agent/api/mod.rs

@@ -50,28 +50,10 @@ fn tools_to_json_str(tools: &[agent_tools::Tool]) -> String {
     format!("[{}]", inner.join(","))
 }
 
-/// Events produced by the streaming API backends.
-/// The runner reads these and decides what to display where.
-pub(crate) enum StreamEvent {
-    /// Content token from the model's response.
-    Content(String),
-    /// Reasoning/thinking token (internal monologue).
-    Reasoning(String),
-    /// Incremental tool call delta (structured, from APIs that support it).
-    ToolCallDelta {
-        index: usize,
-        id: Option<String>,
-        call_type: Option<String>,
-        name: Option<String>,
-        arguments: Option<String>,
-    },
-    /// Token usage stats.
-    Usage(Usage),
-    /// Stream finished.
-    Finished {
-        reason: String,
-    },
-    /// Error from the stream.
+/// One token from the streaming completions API.
+pub(crate) enum StreamToken {
+    Token { text: String, id: u32 },
+    Done { usage: Option<Usage> },
     Error(String),
 }
 
@@ -133,14 +115,14 @@ impl ApiClient {
         (rx, AbortOnDrop(handle))
     }
 
-    /// Start a streaming completion with raw token IDs.
-    /// No message formatting — the caller provides the complete prompt as tokens.
-    pub(crate) fn start_stream_completions(
+    /// Stream a completion with raw token IDs.
+    /// Returns (text, token_id) per token via channel.
+    pub(crate) fn stream_completion(
         &self,
         prompt_tokens: &[u32],
         sampling: SamplingParams,
         priority: Option<i32>,
-    ) -> (mpsc::UnboundedReceiver<StreamEvent>, AbortOnDrop) {
+    ) -> (mpsc::UnboundedReceiver<StreamToken>, AbortOnDrop) {
         let (tx, rx) = mpsc::unbounded_channel();
         let client = self.client.clone();
         let api_key = self.api_key.clone();
@@ -154,7 +136,7 @@ impl ApiClient {
                 &prompt_tokens, &tx, sampling, priority,
             ).await;
             if let Err(e) = result {
-                let _ = tx.send(StreamEvent::Error(e.to_string()));
+                let _ = tx.send(StreamToken::Error(e.to_string()));
             }
         });