fix: prompt caching, conversation length management and token usage display

tests/renderer/navigation/assistantSidebarGuards.test.ts

@@ -35,6 +35,7 @@ describe('assistant sidebar guard rails', () => {
       onToolResult: vi.fn(() => vi.fn()),
       onTitleUpdated: vi.fn(() => vi.fn()),
       onA2UIMessage: vi.fn(() => vi.fn()),
+      onTokenUsage: vi.fn(() => vi.fn()),
       dispatchA2UIAction: vi.fn(),
     } as never;
   });

tests/renderer/navigation/chatSurfaceModeUsageGuards.test.ts

@@ -46,6 +46,7 @@ describe('chat surface mode usage guards', () => {
       onToolResult: vi.fn(() => vi.fn()),
       onTitleUpdated: vi.fn(() => vi.fn()),
       onA2UIMessage: vi.fn(() => vi.fn()),
+      onTokenUsage: vi.fn(() => vi.fn()),
       dispatchA2UIAction: vi.fn(),
     } as never;
   });

tests/renderer/navigation/chatSurfaceUsageGuards.test.ts

@@ -49,6 +49,7 @@ describe('chat surface shared usage guards', () => {
       onToolResult: vi.fn(() => vi.fn()),
       onTitleUpdated: vi.fn(() => vi.fn()),
       onA2UIMessage: vi.fn(() => vi.fn()),
+      onTokenUsage: vi.fn(() => vi.fn()),
       dispatchA2UIAction: vi.fn(),
     } as never;
   });