apidoc template rst_t file does not use utf-8 at reading and writing

[panhaoyu]

Describe the bug
apidoc template rst_t file does not use utf-8 at reading and writing.

To Reproduce

Here is a pytest case, and it will run on my computer, but I think it won't run on non-gbk encoding computers.

def test_utf8_template():
    TEMPLATE = 'There are some Chinese characters below.\n这是一行中文字符\n'

    # Firstly create three temp directories for the template, python source, rst destination directories.
    with tempfile.TemporaryDirectory() as template_dir:
        with tempfile.TemporaryDirectory() as src_dir:
            with tempfile.TemporaryDirectory() as dst_dir:
                # Write a module template rst_t file, use utf-8 encoding.
                # Here I make sure that the template is utf-8.
                with open(os.path.join(template_dir, 'module.rst_t'), mode='w', encoding='utf-8') as f:
                    f.write(TEMPLATE)

                # Create a python file in the python source directory.
                with open(os.path.join(src_dir, 'main.py'), mode='w', encoding='utf-8') as f:
                    f.write('import os\ndef test():\n    print(123)')

                # Use apidoc to generate rst file, and the file generated successfully.
                apidoc_main(['-o', dst_dir, '-t', template_dir, src_dir])
                assert os.path.exists(os.path.join(dst_dir, 'main.rst'))

                # Read the file with utf-8 encoding, but it not work!
                # That's because it is not utf-8 encoding at all!
                with pytest.raises(UnicodeError, match="'utf-8' codec can't decode"):
                    with open(os.path.join(dst_dir, 'main.rst'), mode='r', encoding='utf-8') as f:
                        print(f.read())

                # Read the file with gbk encoding, and it works!
                with open(os.path.join(dst_dir, 'main.rst'), mode='r', encoding='gbk') as f:
                    content = f.read()
                    assert content == TEMPLATE[:-1]

                # That is to say, on my computer, utf-8 is not default encoding.
                # My computer is windows 10 and I'm in China.
                # So, `with open('a.txt', w)` will run as `with open('a.txt', mode='w', encoding='gbk')`.
                # In such systems, we should explicitly specify `encoding='utf-8'` to use utf-8.

                # In fact, when we Chinese learn python file IO,
                # the first thing is to be taught that always use `encoding='utf-8'` in the python world.

Expected behavior
Parse rst_t with non-ascii files.

Your project
https://github.com/panhaoyu/sphinx

Screenshots
The test case is enough I think, or I can provide an image.

Environment info

• OS: Win10
• Python version: 3.7.5
• Sphinx version: 3.x
• Sphinx extensions: no
• Extra tools: no