Suppress rate-limit/quota-exceeded errors from chatAgentError telemetry

[Copilot]

ChatRateLimited errors thrown by the Copilot extension's _provideLanguageModelResponse were surfacing as unhandled errors in the chatAgentError telemetry, despite being expected operational conditions that are already handled gracefully in the UI.

Changes

• src/vs/workbench/api/browser/mainThreadChatAgents2.ts: Skip the publicLogError2('chatAgentError') call when rpcResult.errorDetails?.isRateLimited or isQuotaExceeded is set. Both flags are already populated upstream in extHostChatAgents2.ts based on the error name (ChatRateLimited / ChatQuotaExceeded), so no additional plumbing is needed. The error details continue to flow to the UI layer unchanged.

if (rpcResult?.errorCallstack && !rpcResult.errorDetails?.isRateLimited && !rpcResult.errorDetails?.isQuotaExceeded) {
    this._telemetryService.publicLogError2<ChatAgentErrorEvent, ChatAgentErrorClassification>('chatAgentError', { ... });
}

This corresponds to "fix location 2" from the investigation comment on the issue.

[Copilot]

Pull request overview

This PR reduces noise in workbench telemetry by suppressing chatAgentError telemetry for expected operational failures (rate limiting / quota exceeded) coming from chat agent invocation, while leaving UI error handling unchanged.

Changes:

• Add a guard to skip publicLogError2('chatAgentError') when rpcResult.errorDetails?.isRateLimited or isQuotaExceeded is set.
• Document the rationale inline and link to the tracking issue.

Show a summary per file

File	Description
src/vs/workbench/api/browser/mainThreadChatAgents2.ts	Prevents logging expected rate-limit/quota errors to chatAgentError telemetry while preserving error details for the UI layer.

Copilot's findings

• Files reviewed: 1/1 changed files
• Comments generated: 0