GeezerVison

Fixes #160000.

LZMADecompressor.decompress(b'', max_length=0) works once, then raises LZMAError: Insufficient buffer space on the second call. Same after real input when output is pending internally.

In Modules/_lzmamodule.c, decompress_buf treats LZMA_BUF_ERROR from liblzma as benign only when avail_in == 0 and avail_out > 0. With max_length=0 there is no output space, so that check never fires for the empty call, and the error escapes. The fix treats it as benign when avail_in == 0 or avail_out == 0. That is safe because the avail_out == 0 branch either breaks once the requested length is reached or grows the buffer, so it cannot spin.

Also adds a regression test in Lib/test/test_lzma.py covering the empty case and the after-real-input case, and a NEWS entry.

Verified by building CPython from main and running the test suite: test_lzma passes (125 tests), the issue repro returns b'' on repeated calls, the decompressor stays usable afterwards, and corrupt input still raises LZMAError.

Note: this is my first CPython PR, I will complete the CLA check when prompted.

• Issue: LZMADecompressor.decompress(b'', max_length=0) raises LZMAError on the second consecutive call #160000