Don't include duplicate rows in Gaussian elimination

Duplicate rows in the factorization parity vector row matrix are a double-edged sword. On the one hand, they're expected to hurt the efficiency and potential success of Gaussian elimination; on the other hand, they lead to a smooth perfect-square residue immediately upon combining the duplicates. Hence, don't add any duplicate factorization parity rows to the Gaussian elimination matrix, but, whenever they arise, check immediately for a congruence of squares (and at least retain the original row for Gaussian elimination, if this fails).

Full Changelog: v6.2.2...v6.2.3

sha1sum results:
cd5f43fbed71e72222c590dc9790d8ae248679ef FindAFactor-6.2.3-cp310-cp310-manylinux_2_35_x86_64.whl
5cbe4e8582aecd14615c6e7ce1be39f8e1c8baff FindAFactor-6.2.3-cp312-cp312-manylinux_2_39_x86_64.whl
e5a6729d2c1ee3e50fa7ee7afa30dac5c89cb5fc FindAFactor-6.2.3-cp312-cp312-win_amd64.whl
10381be10ffb80db65adc4dc34bd6aaf4fe4594a FindAFactor-6.2.3-cp313-cp313-macosx_13_0_x86_64.whl
6dd5aac04d06b48ab1934c3ff727b2c1ffa33015 FindAFactor-6.2.3-cp313-cp313-macosx_14_0_arm64.whl
0d9347ae6635a80a7e4b1ee57020c2ff575fa9fa FindAFactor-6.2.3-cp313-cp313-macosx_15_0_arm64.whl
d4076b5db0d716305eb0c9a36480a4c47fdc86bb FindAFactor-6.2.3-cp38-cp38-manylinux_2_31_x86_64.whl
f2a27c285ddcf580cdcc9c57f0addda4c9a690ba findafactor-6.2.3.tar.gz