Continue running next test when async test throws error

Currently async tests end up timing out when they call assert
but don't catch the resulting exception and call done(true).

It seems more natural that when an async test has an assertion
failure, that the harness should catch the exception and take
care of continuing tests.

The alternative that async tests should catch there own errors
has the drawback that specific errors are not ever reported.
E.g. code below is bad.

testAsync(done) {
  try {
    assertEquals('foo', 'bar', 'foo should match bar');
    done();
  } catch  (err) {
    done(true)
  }
}

Bug: 813477
Cq-Include-Trybots: master.tryserver.chromium.linux:closure_compilation
Change-Id: Idbc88f94f6055eec37ab0d90b4b8d423fba6eb3f
Reviewed-on: https://chromium-review.googlesource.com/1088327Reviewed-by: Alexey Baskakov <loyso@chromium.org>
Reviewed-by: Naoki Fukino <fukino@chromium.org>
Commit-Queue: Joel Hockey <joelhockey@chromium.org>
Cr-Commit-Position: refs/heads/master@{#564806}

ui/webui/resources/js/webui_resource_test.js

@@ -205,6 +205,7 @@ function continueTesting(opt_asyncTestFailure) {
     testName = testCases.pop();
     console.log('TEST ' + testName + ' starting...');
     var isAsyncTest = window[testName].length;
+    var testError = false;
     try {
       if (window.setUp)
         window.setUp();
@@ -214,9 +215,11 @@ function continueTesting(opt_asyncTestFailure) {
       console.error('Failure in test ' + testName + '\n' + err);
       console.log(err.stack);
       cleanTestRun = false;
+      testError = true;
     }
-    // Asynchronous tests must manually call continueTesting when complete.
-    if (!isAsyncTest)
+    // Asynchronous tests must manually call continueTesting when complete
+    // unless they throw an exception.
+    if (!isAsyncTest || testError)
       continueTesting();
   } else {
     done = true;