Updated deprecated functions and types

[TheoFABIEN]

This pull request updates deprecated functions and data types to ensure compatibility with recent versions of the dependencies, as the repository does not specify package versions.

Changes include:

• Replaced sklearn.utils.linear_assignment_ with scipy.optimize.linear_sum_assignment (since linear_assignment is deprecated in sklearn).
• Replaced np.int with int (deprecated in NumPy).
• Replaced np.float with float (deprecated in NumPy).

[dyhBUPT]

Hi, thanks for your PR!
After your changes, are the evaluation results consistent with our reported results in paper?

[TheoFABIEN]

With the updated code I was able to successfully run object tracking on the MOT16 dataset.

Then I did some unit testing to double check that my modification for linear assignment gives strictly identical results to the old one.

np.int and np.float were aliases for python types int and float, respectively, and numpy documentation advises to use int and float as replacements.

Unfortunately I did not have enough time to run the evaluation results to compare directly with the paper, but I am confident that they will be the same, given the packages documentation and the tests I did.

If I find some time to do that I could try to run the evaluation results, if you think it is necessary.

[dyhBUPT]

Thanks for your explanation!
To ensure the consistent tracking performance, I will merge this PR after you check the evaluation results on MOT17-val set.