Patchy coverage for `clEnqueueCopyBufferRect`

[kpet]

• No coverage of the case where the provided pitches are zero (see https://github.com/kpet/clvk/pull/753)
• This command is only tested by test_bufferreadwriterect which does not seem to cover all possible region dimensionalities either. We should review what coverage is needed and extend the test accordingly.

[shajder]

@kpet @bashbaug In my evaluation of test coverage for the test_bufferreadwriterect function, I'm assuming that by randomizing the argument values with equal probability, we're likely to cover all necessary scenarios for the function’s dimensional input. This covers all cases of memory allocation sizes, specifically within the limits of (1..CL_DEVICE_MAX_MEM_ALLOC_SIZE) for 3D regions. Please correct me if I am wrong but I can't spot any missing coverage for dimensionalities.

Testing every possible region dimension is unnecessarily exhaustive. Instead, we can focus on key scenarios, such as reducing randomization by selecting special cases. For example, testing could be structured as follows:

• line stretching along the X-axis: (1..maxX, 1, 1),
• plane extending over the X and Y axes: (1..maxX, 1..maxY, 1),
• full 3D volume: (1..maxX, 1..maxY, 1..maxZ), and so on.

Would that be the intended result ?