Findafactor

There seemed to be some instability on the threading of Sieve of Eratosthenes (SoE). Meanwhile, while threading seems to return some small execution time advantage, utilization remains very low for parallel SoE in general. So, threading has been removed from SoE.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v1.6.3...v1.6.4

sha1sum results:
2335ae10181994fb8c7d50f9ef0e8052b57ffcde FindAFactor-1.6.5-cp310-cp310-manylinux_2_35_x86_64.whl
4c521f9968c17a0c9bb0f4a739ba39b0edf5290c FindAFactor-1.6.5-cp312-cp312-manylinux_2_39_x86_64.whl
3f810608c44b0188c03018730054ccb46dc31bde FindAFactor-1.6.5-cp38-cp38-manylinux_2_31_x86_64.whl
dc0f52f9f30ab600101ae51d8c975a5f924d631a findafactor-1.6.5.tar.gz

This release ends inlining of `forward()` and `backward()` functions with `constexpr` arrays, which seems to significantly benefit performance.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v1.6.2...v1.6.3

sha1sum results:
4edf20626bdf78165679064f59ab691ef53a36f8 FindAFactor-1.6.3-cp310-cp310-manylinux_2_35_x86_64.whl
011f346a29537f610e5a6f4fc579c5a7fa7e94ff FindAFactor-1.6.3-cp312-cp312-manylinux_2_39_x86_64.whl
3628fe50fdb69b139d4f420318b3ceac70427198 FindAFactor-1.6.3-cp38-cp38-manylinux_2_31_x86_64.whl
3abd9ae58e8b63ea4fa3d0f918d8b3c7eb6b45de findafactor-1.6.3.tar.gz

Trial division missed 1/64 of factors, due to a bug. (Code has also been refactored for tidiness of the code base.)

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v1.6.1...v1.6.2

sha1sum results:
d996e58767b36c06a280ace7920ca6d3a83fd3fd FindAFactor-1.6.2-cp310-cp310-manylinux_2_35_x86_64.whl
60d6cff4763bb9e40259a244933ccfa39d3ab561 FindAFactor-1.6.2-cp312-cp312-manylinux_2_39_x86_64.whl
e1bacf2ccd5a4f80afe763ceb96b5b35faf5e3b8 FindAFactor-1.6.2-cp38-cp38-manylinux_2_31_x86_64.whl
fb8065396b48d2b1b84f034d4d1c0a7fbf031ae8 findafactor-1.6.2.tar.gz

Trial division did not wait for the asynchronous dispatch to complete before determining if it couldn't find an answer. It also incorrectly blocked further dispatch when a result was returned. These issues have been fixed.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v1.6.0...v1.6.1

sha1sum results:
8b2667a80f94092f7a53c81e019d1b1bde4bc826 FindAFactor-1.6.1-cp310-cp310-manylinux_2_35_x86_64.whl
6de041e107c130d0ab5967bc500449da5067fb44 FindAFactor-1.6.1-cp312-cp312-manylinux_2_39_x86_64.whl
9da582a3010dc96708c2b3750d8092ce35957733 FindAFactor-1.6.1-cp38-cp38-manylinux_2_31_x86_64.whl
0f21036b0da0a3021c574781c13888749a321083 findafactor-1.6.1.tar.gz

All `find_a_factor()` option defaults can now be controlled by environment variables:

- `FINDAFACTOR_USE_CONGRUENCE_OF_SQUARES`
- `FINDAFACTOR_NODE_COUNT`
- `FINDAFACTOR_NODE_ID`
- `FINDAFACTOR_GEAR_FACTORIZATION_LEVEL`
- `FINDAFACTOR_WHEEL_FACTORIZATION_LEVEL`
- `FINDAFACTOR_SMOOTHNESS_BOUND_MULTIPLIER`

These values are overridden by explicit variable values to the function. They have defaults, if the environment variables are not set.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v1.5.0...v1.6.0

sha1sum results:
c31d11955d5acbbc53f81683dcf82a49c1d9d447 FindAFactor-1.6.0-cp310-cp310-manylinux_2_35_x86_64.whl
e48bb14dc4de745fc1d7c76c488ef055794c483f FindAFactor-1.6.0-cp312-cp312-manylinux_2_39_x86_64.whl
16a6e2fc67c26c99678175eff635828ecfcc136e FindAFactor-1.6.0-cp38-cp38-manylinux_2_31_x86_64.whl
6ced54f8e5644f6ea243470d820ee7f05528b38f findafactor-1.6.0.tar.gz

This release adds options to independently vary "wheel" and "gear" factorization levels (explained in the README).

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v1.4.0...v1.5.0

sha1sum results:
26cfcedcd49bc8daca7e7e22455bc292d001a57c FindAFactor-1.5.0-cp310-cp310-manylinux_2_35_x86_64.whl
8f6e7a7489f095dab3e839f9e38c356af5d43923 FindAFactor-1.5.0-cp312-cp312-manylinux_2_39_x86_64.whl
fec1f5b8987a80aa107732b0ee9d3cf1ea309e78 FindAFactor-1.5.0-cp38-cp38-manylinux_2_31_x86_64.whl
b2b17eb78708f6eab462f93ef05ee2b271a5d13a findafactor-1.5.0.tar.gz