ts: basic documentation for utility scripts

Replace the chessboard.png image with the pattern.png image since the chessboard.png image does not have proper squares which is not good for camera calibration accuracy.

docs: prefer # for links generation

- avoid `@ref`
- align with 4.x branch (minimize merge conflicts)

- PCH may not pass compilation flags properly

Add a gif to the homography exercise

Fix the pyramid bug when src*2 < dst

* imgcodecs: tiff: Reduce memory usage to read 16bit image.

* imgcodecs: tiff: Reduce memory usage to read 8bit images

* imgcodecs: tiff: split basic test and full test.

* imgcodecs: tiff: fix to warning C4244

* imgcodecs: tiff: fix to warning C4244

Move marking memory as initialized earlier.

[3.4] Add information about the disparity-to-depth mapping matrix

Add information about the disparity-to-depth mapping matrix. Add more references about other related functions in the calib3d doc.

Report that animated webp is not supported for now

Disabled compiling warnings in case of symbols in cmake for 3.4

build: eliminate uninitialized warnings from GCC12

dnn: eliminate GCC12 warning in total() call

Removes a broken link

Tutorial for Generalized Hough Ballard and Guil Transform

Fix some typos in platforms/, samples/.

Fix angle discretization in Hough transforms

In some situations the last value was missing from the discrete theta
values. Now, the last value is chosen such that it is close to the
user-provided maximum theta, while the distance to pi remains always
at least theta_step/2. This should avoid duplicate detections.

A better way would probably be to use max_theta as is and adjust the
resolution (theta_step) instead, such that the discretization would
always be uniform (in a circular sense) when full angle range is used.