Fix install_opencv in the RoSys docker image to support omnidirectional cameras

[codingpaula]

Currently the docker image doesn't install the opencv requirements like in the pytest action and as specified in the install_opencv.py script. But the dependencies in the script are necessary to e.g. run the tests correctly. When adding the same line as in the pytest action to the Dockerfile, I get the following error:

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/usr/local/lib/python3.12/importlib/__init__.py", line 90, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "<frozen importlib._bootstrap>", line 1387, in _gcd_import
  File "<frozen importlib._bootstrap>", line 1360, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1324, in _find_and_load_unlocked
ModuleNotFoundError: No module named 'install_opencv'

[rodja]

@codingpaula is that fixed with #181?

[codingpaula]

No, #181 just fixes the base image and installing the full project (before running the image resulted in missing the version).

[falkoschindler]

In b43b47c I added checks if omnidir is properly installed. Otherwise we simply ignore corresponding tests. This allows running tests locally and in other environments without the special install_opencv workaround. So this issue can focus on making camera calibration for omnidirectional cameras available to user code based on the RoSys docker image.

[NiklasNeugebauer]

We have now fixed OpenCV to 4.9.0.80 and set the contrib module in the pyproject.
This limits numpy to <2.0 since OpenCV <4.10 has only been compiled with 1.X.
We will update OpenCV as soon as it supports the omnidir module again.

This fixes the issue and omnidir calibration is tested again.