[ETCM-415] Fix random subset of peers where block is broadcasted

[ntallar]

Description

Fixes random subset of peers not working properly

Given a set, Random.shuffle doesn't work properly (kudos to @dmitry-worker for finding this!):

• Random.shuffle((1 to 10).toSet).take(5) == (1 to 10).toSet.take(5), the shuffle does nothing (this seems to happen with more than 4 elements as the HashSet implementation is used)
• Random.shuffle((1 to 10).toSet) == Random.shuffle((1 to 5).toSet), adding more elements doesn't alter the result of the shuffle

This might have lead into bad propagation of blocks, increasing the chances of having forks on our testnet

Proposed Solution

• Converted Set to Seq before shuffling
• Removed broadcast-new-block-hashes config as it's heavily used:
  1. Our current sync implementation relies on this message to have an up-to-date known top, and also triggering the fetching of new blocks
  2. Without sending this message (as when this flag was off) we are not following the Ethereum protocol

[mirkoAlic]

🚀