Improve the performance of at and set, add Mat.atU8, Mat.setU8, etc.

[rainyl]

Releated:

• More efficient at<T>() and set<T>()? #53
• Most efficient way to loop through a Mat #44

simple perf test:

Mat(3840, 2160, CV_8UC1).at: 6937ms
Mat(3840, 2160, CV_8UC1).atU8: 4701ms
Mat(3840, 2160, CV_8UC1).set: 6188ms
Mat(3840, 2160, CV_8UC1).setU8: 3124ms

Maybe faster with Native Assets:

00:10 +305 ~2: test\core\mat_test.dart: Mat at set perf
Mat(3840, 2160, CV_8UC1).at: 6394ms
Mat(3840, 2160, CV_8UC1).atU8: 4512ms
Mat(3840, 2160, CV_8UC1).set: 5258ms
Mat(3840, 2160, CV_8UC1).setU8: 2829ms

at: Mat.atU8, Mat.atI8, Mat.atU16, Mat.atI16, Mat.atI32, Mat.atF32, Mat.atF64

set: Mat.setU8, Mat.setI8, Mat.setU16, Mat.setI16, Mat.setI32, Mat.setF32, Mat.setF64

Mat.set support native types, e.g., Mat.set<cv.U8>()

In previous version, determine the Mat type is very expensive, but considering the mat type may change after creation, so it has to be get from native pointer, which costs lots of time, so for performance sensitive occasions, use atU8/setU8-like API.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 81.44330% with 18 lines in your changes are missing coverage. Please review.

Project coverage is 88.76%. Comparing base (66fa454) to head (f24dab0).

Files	Patch %	Lines
lib/src/core/mat.dart	81.44%	18 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #58      +/-   ##
==========================================
+ Coverage   88.65%   88.76%   +0.11%     
==========================================
  Files          34       34              
  Lines        4777     4789      +12     
==========================================
+ Hits         4235     4251      +16     
+ Misses        542      538       -4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.