fix: update timeout tests to match TaskExecutor error result behavior

[bokelley]

Summary

Fixed timeout handling tests in error-scenarios.test.js to correctly expect error results (success: false) instead of thrown exceptions.

Background

The tests were expecting TaskExecutor to throw TaskTimeoutError exceptions, but the implementation actually catches all exceptions internally and returns standardized error results via the TaskResult interface.

Why CI was passing: These tests are marked as "slow tests" and skipped in CI via skip: process.env.CI (line 17 of the test file). The failures only occurred in local development.

Changes

Updated 4 timeout-related tests:

• ✅ "should timeout on working status after configured limit"
• ✅ "should handle polling timeout during working status"
• ✅ "should handle webhook timeout in submitted tasks" - Fixed async activity issue
• ✅ "should handle zero and negative timeout values"

Test Results

All timeout tests now pass in local development:

// Before (BROKEN):
await assert.rejects(executor.executeTask(...), TaskTimeoutError);

// After (CORRECT):
const result = await executor.executeTask(...);
assert.strictEqual(result.success, false);
assert(result.error.includes('timed out'));

Architecture Verification

✅ Confirmed this is the correct protocol design:

• TaskExecutor.ts line 489: Throws TaskTimeoutError internally
• TaskExecutor.ts line 215: Catches all errors in public API
• TaskExecutor.ts line 766-788: Converts to TaskResult with success: false
• Follows Result pattern (not exception-based error handling)

Related Issues

• These are pre-existing test failures (not related to PR fix(validation): handle framework response wrappers in ADCP schema validation #101)
• Note: Other slow test files may need similar updates (see code review)

Changeset

• 📦 Patch version bump (test-only change)

🤖 Generated with Claude Code