Consistent hashing algorithm leads to uneven distribution of partitions.

Hello, we are using Ignite 2.15 in our project. In practice, we found that the partition assignment method RendezvousAffinityFunction, which is based on consistent hashing, has an issue with uneven distribution. For example, when assigning partitions to 120 nodes, the number of partitions assigned to each node varies significantly. Is there any way to solve this problem?

Alternatively, if our project does not need to consider backups or data migration caused by nodes joining or leaving the cluster, can we simply use a straightforward hash function instead of consistent hashing for partition assignment? If so, what potential issues might arise from this approach?

[RendezvousAffinityFunctionTest.java](https://github.com/user-attachments/files/21968975/RendezvousAffinityFunctionTest.java)

Start by reviewing RendezvousAffinityFunction and the attached RendezvousAffinityFunctionTest.java, focusing on how partitions are assigned across 120 nodes. Compare the observed distribution with the intended assignment guarantees; the issue is complete only once the expected behavior and an agreed remedy are defined and verified.

Written by the indexing model from the issue text.