Inbound peer eviction method - Closes #3721

[mitsuaki-u]

What was the problem?

In the current implementation, when the maximum number of inbound peers are connected and a new peer attempts to connect to the node, we randomly select and remove an ongoing inbound peer connection.

How did I fix it?

In the new protocol, we use the strategy of protecting a number of peers for several distinct characteristics for increased security.

Netgroups: We calculate a randomized value keyedNetgroup based on IP group for each eviction candidate. We protect 4 peers with the smallest keyedNetgroup values. An attacker cannot predict which netgroups will be protected.

• To be started in bucket system milestone

• Peer latency: We protect a number of peers with optimal latency. An attacker cannot manipulate this metric without physically moving nodes closer to the target.

• Useful work: We protect peers who perform useful work, such as peers who have recently sent blocks and transactions. An attacker cannot manipulate this metric without performing useful work.

• Connection time: We protect half of the remaining nodes who have been connected the longest. This precludes attacks that start later.

Review checklist

• The PR resolves Introduce specific eviction method for incoming connections #3721
• All new code is covered with unit tests
• All new code was formatted with Prettier
• Linting passes
• Tests pass
• Commit messages follow the commit guidelines
• Documentation has been added/updated

Addressed.

Addressed.

Addressed.

Addressed

Addressed

Addressed.