Pyqrack

A number of improvements have been made in QBDD simulation:
- `PushStateVector()` has more consistent flooring logic for very small contributions to overall state vector probability.
- Because QBDD gains little advantage from internal parallelism in most cases, internal parallelism in QBDD now defaults to `OFF`.
- If QBDD internal parallelism is off, then `QTensorNetwork` automatically parallelizes QBDD measurement sampling over shots

**Full Changelog**:
https://github.com/unitaryfund/qrack/compare/vm6502q.v9.9.50...vm6502q.v9.9.54
https://github.com/unitaryfund/pyqrack/compare/v1.30.28...v1.30.29

sha1sum results:
d40ee3a18fa6868289d9d725f9d2f644b3fa4767 pyqrack-1.30.29-py3-none-macosx_12_0_x86_64.whl
ac697499c8fa4a84e68381c425059f24e19d2560 pyqrack-1.30.29-py3-none-macosx_13_0_x86_64.whl
69653bfe7d175ee1df08895ae81e16ff7c636ccd pyqrack-1.30.29-py3-none-macosx_14_0_arm64.whl
4125acfaef864b121db805819d6a15e2ba37d2fa pyqrack-1.30.29-py3-none-manylinux_2_31_x86_64.whl
3cd057429e780d58c76c291959b4ba7a9b9d7e4d pyqrack-1.30.29-py3-none-manylinux_2_35_x86_64.whl
05f67c273783f4f9d35ecec80f837bb40fc530be pyqrack-1.30.29-py3-none-manylinux_2_39_x86_64.whl
1cd2b7c25e252ca0dfc6d7245c76c72b47939e87 pyqrack-1.30.29-py3-none-win_amd64.whl
b6c763cbe4c3798fcb1f2f8438865092b52d0403 pyqrack-1.30.29.tar.gz

This release fixes the following:
- an edge case that leads to a segmentation fault when `QStabilizerHybrid` is used in the presence of 0 available OpenCL/CUDA devices,
- potentially inconsistent defaults for qubit-limit and rounding-parameter environment variables,
- and rounding neighborhood for separability checks,if we check a two-state qubit for eigenstate norm equality: "machine epsilon" in C++ is the difference between 1.0 and the next greater representable number; half this value is the difference between 1.0 and the next lower representable number; half this value is used for (L2-norm) rounding neighborhood for equality with 1.0 or 0.0 in two-state Hilbert spaces (which are bounded to that range and must normalize to 1.0).

**Full Changelog**:
https://github.com/unitaryfund/qrack/compare/vm6502qv9.9.46...vm6502q.v9.9.50
https://github.com/unitaryfund/pyqrack/compare/v1.30.26...v1.30.28

sha1sum results:
36d3b47b79a5ead25f36cf30a622acfe0098167f pyqrack-1.30.28-py3-none-macosx_12_0_x86_64.whl
225bf79d0561bfa9f66ac879ee1ea126889ec7df pyqrack-1.30.28-py3-none-macosx_13_0_x86_64.whl
f63661664b7d668a1d80606fd0096958b2641187 pyqrack-1.30.28-py3-none-macosx_14_0_arm64.whl
325061786de3436c136ac161388ab56cdf7ce176 pyqrack-1.30.28-py3-none-manylinux_2_31_x86_64.whl
a325a192cebd632ca836f3965117bd3635b2b1d0 pyqrack-1.30.28-py3-none-manylinux_2_35_x86_64.whl
f8445f8917e390d4e3a181c541e0cda1b3aff17b pyqrack-1.30.28-py3-none-manylinux_2_39_x86_64.whl
6cf1a0e396c5c059b9f1e986b98f3ebdf199f70a pyqrack-1.30.28-py3-none-win_amd64.whl
35e413d6e6f906ff06b9afc3f14f6218b00baa5a pyqrack-1.30.28.tar.gz

This release fixes an edge case that leads to a segmentation fault, when `QTensorNetwork` is used in the presence of 0 available OpenCL/CUDA devices.

**Full Changelog**:
https://github.com/unitaryfund/qrack/compare/vm6502q.v9.9.42...vm6502qv9.9.46
https://github.com/unitaryfund/pyqrack/compare/v1.30.25...v1.30.26

sha1sum results:
1a21b667a26479cbf1af957a7f79d6e0138c8d30 pyqrack-1.30.26-py3-none-macosx_12_0_x86_64.whl
2ccc91deb45ab321da8b0137787b5fee4e2a1ce6 pyqrack-1.30.26-py3-none-macosx_13_0_x86_64.whl
3cde8a9f9f9484ec1c898149607859ec0033aa6f pyqrack-1.30.26-py3-none-macosx_14_0_arm64.whl
0025c24886a4dd2e78ecb0d2fdafe20f3121ee52 pyqrack-1.30.26-py3-none-manylinux_2_31_x86_64.whl
2b22970e52ef7cc9ce80ed402dc6d51e260a339a pyqrack-1.30.26-py3-none-manylinux_2_35_x86_64.whl
394570187a5c16374e4a496bba84b6ed1540ea62 pyqrack-1.30.26-py3-none-manylinux_2_39_x86_64.whl
fddbadf15c690f988cd887db779d0be309177765 pyqrack-1.30.26-py3-none-win_amd64.whl
c018f087b86a9a72f3c3267a270d853f3a34af49 pyqrack-1.30.26.tar.gz

A potential deadlock in QBDD Kronecker products has been fixed. Beyond that, several (non-breaking) function prototype changes have been made in accordance with standards for style.

**Full Changelog**:
https://github.com/unitaryfund/qrack/compare/vm6502q.v9.9.37...vm6502q.v9.9.42
https://github.com/unitaryfund/pyqrack/compare/v1.30.24...v1.30.25

sha1sum results:
df7cb461df2ab3a2e69372ae27a846399adf3ee0 pyqrack-1.30.25-py3-none-macosx_12_0_x86_64.whl
57257c9a26f6d8e0017ae6ec8312427c4c4c1320 pyqrack-1.30.25-py3-none-macosx_13_0_x86_64.whl
36d8e9c31ca13aa40aebf15b8f93b1bbfc236596 pyqrack-1.30.25-py3-none-macosx_14_0_arm64.whl
9a46644ebd3624b6c7eabf3b82d61fd34c1c6fb2 pyqrack-1.30.25-py3-none-manylinux_2_31_x86_64.whl
d1d0f767b8d29c39683ad6d7c8a2b9500782205b pyqrack-1.30.25-py3-none-manylinux_2_35_x86_64.whl
a5dc1c31ff6624a1c0542807fdcc86dc4f576149 pyqrack-1.30.25-py3-none-manylinux_2_39_x86_64.whl
2dda4944582870aa8b4b80065baa219990bca35a pyqrack-1.30.25-py3-none-win_amd64.whl
e70743b45ee9e8a6cff6db3de2374189bd7953de pyqrack-1.30.25.tar.gz

QBDD and stabilizer methods have been optimized. An incorrect initialization of a default `QEngine` property was also fixed in the default constructor (though the incorrect initialization likely had no consequence).

**Full Changelog**:
https://github.com/unitaryfund/qrack/compare/vm6502q.v9.9.33...vm6502q.v9.9.37
https://github.com/unitaryfund/pyqrack/compare/v1.30.23...v1.30.24

sha1sum results:
5c87a32940566928e559e32104dd326b41add384 pyqrack-1.30.24-py3-none-macosx_12_0_x86_64.whl
c27b7239ff36296fcc858d33a1dbdb9775506711 pyqrack-1.30.24-py3-none-macosx_13_0_x86_64.whl
1731d517468d45ca1e05ea56b3196e77105f62a4 pyqrack-1.30.24-py3-none-macosx_14_0_arm64.whl
8560d24506aeb8d43efb58fdf41545995a582ddf pyqrack-1.30.24-py3-none-manylinux_2_31_x86_64.whl
de19ae67645b40afac9b3e28c66c417867fe381d pyqrack-1.30.24-py3-none-manylinux_2_35_x86_64.whl
d9dd179294ccf615534ce658b2b249240c6eaf59 pyqrack-1.30.24-py3-none-manylinux_2_39_x86_64.whl
77504c0efa42328d996ff7fcfb4ace171cf04db6 pyqrack-1.30.24-py3-none-win_amd64.whl
6eca84f19dd3d2be62f913ef97099f24427a281a pyqrack-1.30.24.tar.gz

QBdt` and `BigInteger` (alias `bitCapInt`) have been debugged. Additionally, one should see drastically improved performance with `QBdt` (despite significantly reduced parallelism utilization). Systemically, on GNU/Linux, the `bitCapInt` type now uses 128-bit "words" rather than 64-bit words, which should lead to some performance improvement.

Note that previous releases were almost universally built with capacity for 4095 qubits in a single simulator instance; in practicality, little besides stabilizer simulation can take advantage of the vast majority of this addressing capacity. This release limits to 127 qubits in a single simulator instance, but this is a "feasible" width for Qrack approximation techniques, and performance benefits from the change.

**Full Changelog**:
https://github.com/unitaryfund/qrack/compare/vm6502q.v9.9.28...vm6502q.v9.9.33
https://github.com/unitaryfund/pyqrack/compare/v1.30.22...v1.30.23

sha1sum results:
7d56a837c69f0ef15154d540295c09700a5a06eb pyqrack-1.30.23-py3-none-macosx_12_0_x86_64.whl
d581d679f0230b4ebc6bd42edd157cad12c2b83e pyqrack-1.30.23-py3-none-macosx_13_0_x86_64.whl
9c843975e02018239239735129ca52ba2d4708be pyqrack-1.30.23-py3-none-macosx_14_0_arm64.whl
0d6b638b3cb56e3544af0ddfa9ab5480b814e004 pyqrack-1.30.23-py3-none-manylinux_2_31_x86_64.whl
99bf601b21357239b37cffa4366606fa88b14202 pyqrack-1.30.23-py3-none-manylinux_2_35_x86_64.whl
f074b480037da89158ec87a2c8d063d69589f9ad pyqrack-1.30.23-py3-none-manylinux_2_39_x86_64.whl
64261eb452e35173b2fbd186138b7d8ec4d7f8dc pyqrack-1.30.23-py3-none-win_amd64.whl
aeab93744971995f2caad91ba440475486527eab pyqrack-1.30.23.tar.gz