agent: Fix issues with usage display sometimes showing initially fetched usage (#33125)

Having `Thread::last_usage` as an override of the initially fetched
usage could cause the initial usage to be displayed when the current
thread is empty or in text threads. Fix is to just store last usage info
in `UserStore` and not have these overrides

Release Notes:

- Agent: Fixed request usage display to always include the most recently
known usage - there were some cases where it would show the initially
requested usage.

crates/agent/src/thread.rs

@@ -7,6 +7,7 @@ use agent_settings::{AgentProfileId, AgentSettings, CompletionMode};
 use anyhow::{Result, anyhow};
 use assistant_tool::{ActionLog, AnyToolCard, Tool, ToolWorkingSet};
 use chrono::{DateTime, Utc};
+use client::{ModelRequestUsage, RequestUsage};
 use collections::HashMap;
 use editor::display_map::CreaseMetadata;
 use feature_flags::{self, FeatureFlagAppExt};
@@ -22,8 +23,8 @@ use language_model::{
     LanguageModelId, LanguageModelKnownError, LanguageModelRegistry, LanguageModelRequest,
     LanguageModelRequestMessage, LanguageModelRequestTool, LanguageModelToolResult,
     LanguageModelToolResultContent, LanguageModelToolUseId, MessageContent,
-    ModelRequestLimitReachedError, PaymentRequiredError, RequestUsage, Role, SelectedModel,
-    StopReason, TokenUsage,
+    ModelRequestLimitReachedError, PaymentRequiredError, Role, SelectedModel, StopReason,
+    TokenUsage,
 };
 use postage::stream::Stream as _;
 use project::Project;
@@ -38,7 +39,7 @@ use ui::Window;
 use util::{ResultExt as _, post_inc};
 
 use uuid::Uuid;
-use zed_llm_client::{CompletionIntent, CompletionRequestStatus};
+use zed_llm_client::{CompletionIntent, CompletionRequestStatus, UsageLimit};
 
 use crate::ThreadStore;
 use crate::agent_profile::AgentProfile;
@@ -350,7 +351,6 @@ pub struct Thread {
     request_token_usage: Vec<TokenUsage>,
     cumulative_token_usage: TokenUsage,
     exceeded_window_error: Option<ExceededWindowError>,
-    last_usage: Option<RequestUsage>,
     tool_use_limit_reached: bool,
     feedback: Option<ThreadFeedback>,
     message_feedback: HashMap<MessageId, ThreadFeedback>,
@@ -443,7 +443,6 @@ impl Thread {
             request_token_usage: Vec::new(),
             cumulative_token_usage: TokenUsage::default(),
             exceeded_window_error: None,
-            last_usage: None,
             tool_use_limit_reached: false,
             feedback: None,
             message_feedback: HashMap::default(),
@@ -568,7 +567,6 @@ impl Thread {
             request_token_usage: serialized.request_token_usage,
             cumulative_token_usage: serialized.cumulative_token_usage,
             exceeded_window_error: None,
-            last_usage: None,
             tool_use_limit_reached: serialized.tool_use_limit_reached,
             feedback: None,
             message_feedback: HashMap::default(),
@@ -875,10 +873,6 @@ impl Thread {
             .unwrap_or(false)
     }
 
-    pub fn last_usage(&self) -> Option<RequestUsage> {
-        self.last_usage
-    }
-
     pub fn tool_use_limit_reached(&self) -> bool {
         self.tool_use_limit_reached
     }
@@ -1658,9 +1652,7 @@ impl Thread {
                                         CompletionRequestStatus::UsageUpdated {
                                             amount, limit
                                         } => {
-                                            let usage = RequestUsage { limit, amount: amount as i32 };
-
-                                            thread.last_usage = Some(usage);
+                                            thread.update_model_request_usage(amount as u32, limit, cx);
                                         }
                                         CompletionRequestStatus::ToolUseLimitReached => {
                                             thread.tool_use_limit_reached = true;
@@ -1871,11 +1863,8 @@ impl Thread {
                         LanguageModelCompletionEvent::StatusUpdate(
                             CompletionRequestStatus::UsageUpdated { amount, limit },
                         ) => {
-                            this.update(cx, |thread, _cx| {
-                                thread.last_usage = Some(RequestUsage {
-                                    limit,
-                                    amount: amount as i32,
-                                });
+                            this.update(cx, |thread, cx| {
+                                thread.update_model_request_usage(amount as u32, limit, cx);
                             })?;
                             continue;
                         }
@@ -2757,6 +2746,20 @@ impl Thread {
         }
     }
 
+    fn update_model_request_usage(&self, amount: u32, limit: UsageLimit, cx: &mut Context<Self>) {
+        self.project.update(cx, |project, cx| {
+            project.user_store().update(cx, |user_store, cx| {
+                user_store.update_model_request_usage(
+                    ModelRequestUsage(RequestUsage {
+                        amount: amount as i32,
+                        limit,
+                    }),
+                    cx,
+                )
+            })
+        });
+    }
+
     pub fn deny_tool_use(
         &mut self,
         tool_use_id: LanguageModelToolUseId,