bpo-1635741: Port _lzma module to multiphase initialization

[corona10]

https://bugs.python.org/issue1635741

[corona10]

I ran the several leak test also for subinterp.
No leak was found on this change.

class SubinterpThreadingTests(unittest.TestCase):
    def test_lzma_subinterp(self):
        code = textwrap.dedent(r"""
            import os
            from lzma import LZMACompressor, LZMADecompressor, LZMAError
            lzc = LZMACompressor()
            lzd = LZMADecompressor()
            def noop(*args):
                pass

            os.register_at_fork(after_in_child=noop)
        """)
        ret = support.run_in_subinterp(code)

[corona10]

@shihai1991 Can you please take a look?

[shihai1991]

thanks, LGTM.

[corona10]

@vstinner Please take a look :)

[corona10]

./python.exe -m test test_capi -R 3:3
0:00:00 load avg: 1.34 Run tests sequentially
0:00:00 load avg: 1.34 [1/1] test_capi
beginning 6 repetitions
123456
......
test_capi passed in 34.3 sec

== Tests result: SUCCESS ==

1 test OK.

Total duration: 34.3 sec
Tests result: SUCCESS

with patch (https://bugs.python.org/issue40149#msg365560)

./python.exe -m test test_threading -R 3:3
0:00:00 load avg: 1.84 Run tests sequentially
0:00:00 load avg: 1.84 [1/1] test_threading
beginning 6 repetitions
123456
......
test_threading passed in 1 min 11 sec

== Tests result: SUCCESS ==

1 test OK.

Total duration: 1 min 11 sec
Tests result: SUCCESS

[vstinner]

LGTM. Just two minor coding style remarks.