3.x: stop throwing log message on zero token nodes

[dkropachev]

Currently driver logs warning message on every metadata update for each zero token node:

9932   [main] WARN  com.datastax.driver.core.ControlConnection - Found invalid row in system.peers: [peer=/127.0.1.3, missing native_transport_address, missing native_transport_port, missing native_transport_port_ssl, tokens=null]. This is likely a gossip or snitch issue, this host will be ignored.

It produce bad UX, we need to stop doing that.

[dkropachev]

It was missed when we where working on #317 for 3.x

[Bouncheck]

Isn't it logged only when option is not enabled? Looks to me like it will be logged only if it is supposed to be considered invalid

java-driver/driver-core/src/main/java/com/datastax/driver/core/ControlConnection.java

Lines 1004 to 1023 in 1ca5689

	if (EXTENDED_PEER_CHECK) {
	isValid &=
	peerRow.getColumnDefinitions().contains("data_center")
	&& !peerRow.isNull("data_center")
	&& peerRow.getColumnDefinitions().contains("rack")
	&& !peerRow.isNull("rack")
	&& peerRow.getColumnDefinitions().contains("tokens")
	&& (!peerRow.isNull("tokens")
	|| cluster
	.configuration
	.getQueryOptions()
	.shouldConsiderZeroTokenNodesValidPeers());
	}
	if (!isValid && logIfInvalid)
	logger.warn(
	"Found invalid row in system.peers: {}. "
	+ "This is likely a gossip or snitch issue, this host will be ignored.",
	formatInvalidPeer(peerRow));
	return isValid;
	}

[dkropachev]

Say you have a cluster with zero token nodes, but you don't want your driver connect to them.
Then shouldConsiderZeroTokenNodesValidPeers is false, but you are getting log messages on every metadata update.
It is a bad UX, we should silently skip them.

[dkropachev]

Fixed - #519