0.9.14

 - build: CUDA 13 build

0.9.13

 - ref: do not import torch from dcnum, actually has no effect
 - setup: bump dcnum to 0.29.1

0.9.12

 - fix: import torch from dcnum to make sure variables are set correctly

0.9.11

 - build: update PyInstaller hook for dcnum

0.9.10

 - enh: put `main` methods in `__name__ == "__main__"` guard
 - enh: check for cuda availability instead of letting dcnum decide
 - setup: bump dcnum to 0.29.0
 - setup: update pyinstaller hooks

0.9.9

 - fix: reduce idle CPU usage of GUI and sail around memory leak in dcnum
 - setup: bump dcnum to 0.28.3 (internal basin sorted first)

0.9.8

 - ref: replace `len(os.sched_getaffinity(0))` with `dcnum.common.cpu_count()`

0.9.7

 - reg: deselect GPU segmenter when not available
 - ref: replace `mp.cpu_count()` with `len(os.sched_getaffinity(0))`
 - setup: bump dcnum to 0.28.1 (fix broken extractor for invalid events)

0.9.6

 - enh: set "use GPU" as default
 - enh: compute Haralick texture features by default
 - setup: bump dcnum to 0.28.0

0.9.5

 - docs: list more software in Help/Software menu
 - build: use CUDA 12.6 to reduce output file size