Implement methods to detect and eliminate outliers

[opatiny]

They are at least these two possibilities to remove some of the outliers:

1) crosscheck: best match for a-> should be the same for b->a https://docs.opencv.org/4.x/dc/dc3/tutorial_py_matcher.html

2) Lowe's ratio test: the best match $p_1$ is at least two times better that the second match $p_2$ $\frac{d(q,p_1)}{d(q,p_2)} < \frac{1}{2}$

[opatiny]

Crosscheck was implemented in #199

[opatiny]

Lowe's ratio test will necessitate implementing a function that returns the N best matches for each keypoint.

[opatiny]

A basic function to eliminate outliers was implemented (filterEucledeanDistance), but it can only be used when the rough position of the destination is known. It only keeps the best match from source to a given destination keypoint, "the better" meaning the one which euclidean distance is the smallest.

Implemented in c590cdb27e1d13d56833f4dbb0e3bd05f38dbfab