Switch from GetDIBits to CreateDIBSection for screen capture on Windows

[halldorfannar]

Changes proposed in this PR

Switch from GetDIBits to CreateDIBSection for screen capture on Windows. This is more friendly in terms of how we utilize memory but did not yield performance improvements.

This PR includes a new benchmark, src/tests/bench_grab_windows.py, that was used to measure the performance before and after the changes on two different Windows machines.

My measurements shows similar performance when comparing these two approaches.

Test results on a Qualcomm arm64 + Qualcomm Adreno X1-85 GPU:

Region	Original (GetDIBits)	New (CreateDIBSection)
Full (2304x1536)	35.72ms (28 FPS)	34.49ms (29 FPS)
100x100	13.34ms (75 FPS)	14.67ms (68 FPS)
640x480	16.43ms (61 FPS)	14.18ms (71 FPS)

Test results on a Intel x86_64 + NVIDIA A2000 GPU:

Region	Original (GetDIBits)	New (CreateDIBSection)
Full (3840x2160)	81.64ms (12 FPS)	82.36ms (12 FPS)
100x100	8.33ms (120 FPS)	8.33ms (120 FPS)
640x480	8.33ms (120 FPS)	8.33ms (120 FPS)

That being said, this new approach involves less memory churn and after testing these changes against issue #268, I was unable to reproduce the issue reported there. Of course, the final verdict will be in the hands of the user who reported the original issue - but closing the issue here as fixed is the right action, so that users will be encouraged to test this on their machines.

Fixes #449, fixes #268

• Tests added/updated
• Documentation updated
• Changelog entry added
• ./check.sh passed

[BoboTiG]

Wow, that's perfect! Thanks a lot @halldorfannar 🥂