interfaces into a single place. The interfaces themselves are located in
sub-modules, as described below.

Note that the file `__init__.py` in the TensorFlow source code tree is actually
only a placeholder to enable test cases to run. The TensorFlow build replaces

    testsuites.attrib = suite_attrib
    testsuite.attrib = suite_attrib
    indent_xml(testsuites)

  tree = ElemTree.ElementTree(testsuites)
  file_path = os.path.join(output_path)
  with open(file_path, 'wb') as f:
    f.write(b'<?xml version="1.0"?>\n')
    tree.write(f)
    if verbose:
      print(f'\nWrote XML with Bazel invocation results to {file_path}')

)

_T = TypeVar("_T")


@asynccontextmanager
async def contextmanager_in_threadpool(
    cm: ContextManager[_T],
) -> AsyncGenerator[_T, None]:
    # blocking __exit__ from running waiting on a free thread
    # can create race conditions/deadlocks if the context manager itself
    # has its own internal pool (e.g. a database connection pool)
    # to avoid this we let __exit__ run without a capacity limit