Added fisheye camera (equidistant model) support and fixed some minor errors

[mgrova]

This pull request adds support for cameras with equidistant or fisheye distortion models. To do this, a method has been created that allows the camera model used to be identified and, with this information, a method for projecting the points or another method to be used.

In addition, the following have been fixed/modified:

• Small typo in the README.
• When clicking on the rectifyimage checkbox, the point cloud disappeared and did not reappear, and the following error was displayed on the screen:

Traceback (most recent call last):
  File "/home/mmontes/Developments/ros2_calib/.venv/lib/python3.10/site-packages/ros2_calib/calibration_widget.py", line 210, in toggle_rectification
    self.display_image()  # Refresh the display
  File "/home/mmontes/Developments/ros2_calib/.venv/lib/python3.10/site-packages/ros2_calib/calibration_widget.py", line 844, in display_image
    self.project_pointcloud()
  File "/home/mmontes/Developments/ros2_calib/.venv/lib/python3.10/site-packages/ros2_calib/calibration_widget.py", line 898, in project_pointcloud
    if self.point_cloud_item is not None and self.point_cloud_item.scene():
RuntimeError: Internal C++ object (PointCloudItem) already deleted.

To fix this, this snippet has been added.

• Use the function ‘read_all_messages_optimized’ instead of ‘read_synchronized_image_cloud’ to obtain the selectable frames. Although ‘read_synchronized_image_cloud’ is much more efficient, it only takes the first N synchronized frames. However, I think it is better to obtain N evenly spaced frames throughout the entire bag, as this allows for different points of view and selection of the most appropriate one.

I would be delighted to receive any comments or modifications you may have :)

Bests,
Marco.