Better recovery after reset on high pool load

[rPraml]

There is an issue, if there is high load on the pool.

in the obtainConnection there is code, that prefers "waitingThreads" with the next connection, that comes back in the pool.
what can happen now:

• if the connection pool is near the limit (here we use maxSize=1) and we hold the connection, while the heartbeat runs, the heartbeat might reset the pool.
• waiting threads are preferred to get the next connection, that is returned to the pool in _obtainConnectionWaitLoop
• but there is no connection, that will get back to the loop (pool was reseted)

I've simplified the test, it might be, that this test is flaky now, but it better demonstrates the issue.

Without this fix, we have one successful execution (see first commit), after the fix, all executions are successful.