Non-overlapping functionality with some other adjustments

[TabassumNova]

This repository was incredibly helpful during my master's thesis. In our use case, we worked with 6 cameras that did not have overlapping views. To address this, I used an icosahedron as the calibration object. The dataset for this project is available here

In this pull request, I have added the following:

1. A hand-eye calibration method to address the non-overlapping camera scenario.
2. An iterative approach for calculating intrinsic parameters, which proves more robust for large datasets and reduces the need to manually select suitable images for intrinsic calibration.
3. Functionality to reject outlier poses from further calculations.

The complete visualization for initial guess for camera extrinsic calibration and the final estimation is available here