feat(net): optimize disconnectRandom by tracking block receive time

[xxo1shine]

What does this PR do?

Optimize disconnectRandom by tracking block receive time peer

• Add blockRcvTime/blockRcvTimeCmp fields to PeerConnection to track when a peer last delivered a valid block
• Set blockRcvTime in BlockMsgHandler after each block is received
• Fix lastInteractiveTime update in InventoryMsgHandler: only update for block inventories above current head block num, preventing attackers from forging activity via stale block hashes
• Add getRandomDisconnectionPeers() to ResilienceService: narrows the disconnect candidate pool to the oldest half by blockRcvTime, so peers that recently delivered blocks are protected from random eviction

Fixes #6504

Why are these changes required?

This PR has been tested by:

• Unit Tests
• Manual Testing

Follow up

Extra details

[warku123]

[QUEWSTION]Why does blockRcvTimeCmp live in PeerConnection?

Its only use is inside ResilienceService.getRandomDisconnectionPeers() to snapshot blockRcvTime before sorting, so the comparator reads a stable value even if another thread updates blockRcvTime mid-sort. That's a valid concern, but it's an implementation detail of one method in one service — it doesn't describe any property of the peer itself.

A local snapshot map achieves the same safety without polluting the peer domain object:

private List<PeerConnection> getRandomDisconnectionPeers(List<PeerConnection> peers) {
    Map<PeerConnection, Long> snapshot = new IdentityHashMap<>();
    peers.forEach(p -> snapshot.put(p, p.getBlockRcvTime()));
    List<PeerConnection> sorted = new ArrayList<>(peers);
    sorted.sort(Comparator.comparingLong(snapshot::get));
    return sorted.subList(0, sorted.size() / 2);
}

Is there a reason this needs to be a field rather than a method-local variable?