Prefer images with less polls

In the previous versions, the shown images were selected randomly (uniformly distributed). Now, we use an betavariate distribution: Images with less polls are preferable been shown to the user. This should help to faster achieve a certain number of polls and an acceptable median value.

Furthermore, some libraries were updates due to security issues.

DISCOLI-Version

This version has been presented at the DISCOLI-workshop 2023:

Animals in the Wild: Using Crowdsourcing to enhance the labelling of Camera Trap Images

-> https://discoli-workshop.github.io/2023/program.html