Add test runner script for cobalt_browsertests on Linux starboard (#8582)

Test runner script to manage the test process life cycle.

Starboard cannot launch a new process as cobalt runs in single process
mode.
In the browser test, each test case must own its process, so we have to
run one test case each time.
The python test wrapper helps manage the test processes.

```
python3 cobalt/testing/browser_tests/run_browser_tests.py out/linux-x64x11_devel/cobalt_browsertests "ContentMainRunnerImplBrowserTest*"
...
[       OK ] ContentMainRunnerImplBrowserTest.StartupSequence (122 ms)
[----------] 1 test from ContentMainRunnerImplBrowserTest (122 ms total)

[----------] Global test environment tear-down
[==========] 1 test from 1 test suite ran. (122 ms total)
[  PASSED  ] 1 test.

========================================
Total: 1, Passed: 1, Failed: 0

All tests PASSED.
```

Issue: 433354983

Author: linxinan-yt

cobalt/testing/browser_tests/run_browser_tests.py

@@ -0,0 +1,144 @@
+#!/usr/bin/env python3
+"""Helper script to run Cobalt browser tests.
+
+Cobalt is designed to run in single process mode on TV devices. Its
+browsertests is following it with --single-process-test as well.
+To avoid regressions crashed the process, the test process management
+is left in this python script, which list each single test case
+and iterate them.
+"""
+import subprocess
+import sys
+import os
+
+
+def main():
+  if len(sys.argv) < 2:
+    print("Usage: python3 run_browser_tests.py <path_to_executable> "
+          "[gtest_filter]")
+    sys.exit(1)
+
+  binary_path = sys.argv[1]
+  extra_filter = sys.argv[2] if len(sys.argv) > 2 else "*"
+
+  if not os.path.isfile(binary_path):
+    print(f"Error: Binary not found at {binary_path}", file=sys.stderr)
+    sys.exit(1)
+
+  # 1. List all tests
+  print(f"Listing tests from {binary_path}...")
+  try:
+    output = subprocess.check_output(
+        [binary_path, "--gtest_list_tests", f"--gtest_filter={extra_filter}"],
+        text=True)
+  except subprocess.CalledProcessError as e:
+    print("Failed to list tests.", file=sys.stderr)
+    sys.exit(e.returncode)
+
+  tests = []
+  current_suite = None
+  for line in output.splitlines():
+    # Clean up the line
+    line = line.strip()
+
+    # Skip empty lines or logs (logs usually start with [pid:...)
+    if not line or line.startswith("["):
+      continue
+
+    # Remove comments (starting with #)
+    line = line.split("#")[0].strip()
+
+    if not line:
+      continue
+
+    if line.endswith("."):
+      current_suite = line
+    else:
+      # It's a test name
+      if current_suite:
+        full_test_name = f"{current_suite}{line}"
+        if "DISABLED_" not in full_test_name:
+          tests.append(full_test_name)
+      else:
+        print(
+            f"Warning: Found test '{line}' without a suite. Skipping.",
+            file=sys.stderr)
+
+  if not tests:
+    print("No tests found matching the filter.")
+    sys.exit(0)
+
+  # Sort tests to ensure PRE_ tests run before their main tests.
+  # Logic: PRE_PRE_Test -> PRE_Test -> Test
+  def get_sort_key(test_name):
+    # test_name is Suite.Test
+    parts = test_name.split(".", 1)
+    if len(parts) != 2:
+      return (test_name, 0)
+    suite, case = parts
+
+    pre_count = 0
+    while case.startswith("PRE_"):
+      case = case[4:]
+      pre_count += 1
+
+    # Sort by base name (Suite.TestBase), then by pre_count DESCENDING
+    return (f"{suite}.{case}", -pre_count)
+
+  tests.sort(key=get_sort_key)
+
+  print(f"Found {len(tests)} tests. Running them one by one...\n")
+
+  # 2. Run each test in a fresh process
+  failed_tests = []
+  passed_count = 0
+
+  for i, test in enumerate(tests):
+    print(f"[{i+1}/{len(tests)}] Running {test}...")
+    try:
+      # We pass the filter to run EXACTLY this test case.
+      # Using --gtest_filter=ExactTestName
+      # Add standard Cobalt flags:
+      #   --no-sandbox --single-process --no-zygote --ozone-platform=starboard
+      cmd = [
+          binary_path,
+          f"--gtest_filter={test}",
+          "--single-process-tests",
+          "--no-sandbox",
+          "--single-process",
+          "--no-zygote",
+          "--ozone-platform=starboard",
+      ]
+
+      # Run the test and let it print to stdout/stderr
+      retcode = subprocess.run(cmd, check=False).returncode
+
+      if retcode != 0:
+        print(f"FAILED: {test} (Exit code: {retcode})")
+        failed_tests.append(test)
+      else:
+        passed_count += 1
+
+    except KeyboardInterrupt:
+      print("\nAborted by user.")
+      sys.exit(130)
+    except Exception as e:  # pylint: disable=broad-exception-caught
+      print(f"Error running {test}: {e}")
+      failed_tests.append(test)
+
+  print("\n" + "=" * 40)
+  print(f"Total: {len(tests)}, Passed: {passed_count}, "
+        f"Failed: {len(failed_tests)}")
+
+  if failed_tests:
+    print("\nFailed Tests:")
+    for t in failed_tests:
+      print(f"  {t}")
+    sys.exit(1)
+  else:
+    print("\nAll tests PASSED.")
+    sys.exit(0)
+
+
+if __name__ == "__main__":
+  main()