Fast startup: only retokenize tail of conversation log

restore_from_log reads the full log but walks backwards from the tail,
retokenizing each node as it goes. Stops when conversation budget is
full. Only the nodes that fit get pushed into context.

Added AstNode::retokenize() — recomputes token_ids on all leaves
after deserialization (serde skip means they're empty).

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

src/agent/mod.rs

@@ -568,7 +568,7 @@ impl Agent {
     }
 
     pub async fn restore_from_log(&self) -> bool {
-        let nodes = {
+        let all_nodes = {
             let ctx = self.context.lock().await;
             match &ctx.conversation_log {
                 Some(log) => match log.read_nodes(64 * 1024 * 1024) {
@@ -579,17 +579,35 @@ impl Agent {
             }
         };
 
+        // Walk backwards from the tail, retokenize, stop at budget
+        let budget = context::context_budget_tokens();
+        let fixed = {
+            let ctx = self.context.lock().await;
+            ctx.system().iter().chain(ctx.identity().iter())
+                .map(|n| n.tokens()).sum::<usize>()
+        };
+        let conv_budget = budget.saturating_sub(fixed);
+
+        let mut kept = Vec::new();
+        let mut total = 0;
+        for node in all_nodes.into_iter().rev() {
+            let node = node.retokenize();
+            let tok = node.tokens();
+            if total + tok > conv_budget && !kept.is_empty() { break; }
+            total += tok;
+            kept.push(node);
+        }
+        kept.reverse();
+
         {
             let mut ctx = self.context.lock().await;
             ctx.clear(Section::Conversation);
-            // Push without logging — these are already in the log
-            for node in nodes {
+            for node in kept {
                 ctx.push_no_log(Section::Conversation, node);
             }
         }
         self.compact().await;
-        let mut st = self.state.lock().await;
-        st.last_prompt_tokens = self.context.lock().await.tokens() as u32;
+        self.state.lock().await.last_prompt_tokens = self.context.lock().await.tokens() as u32;
         true
     }