Still a bit crude, needs more error checking.

Off by default, of course. :) Configurable per-connection,
simply send "activate-stale-reads".

Fixes EOS-2072, thanks to Herve Rousseau for the bug report

This should make the cluster more stable (ie no spurious timeouts)
under heavy load.

At least 3x throughput increase on my local machine. ^^

Still some rough edges.. Need to also refactor TestNode
in test-utils to allow easy testing of bulkload and standalone modes.