Missing null check for cluster connections

[szehetner]

This adds a null check in Dispose(ErrorHandler errorHandler, IDisposable disposable), similar to the other CloseHelper.Dispose methods.

Background:
When I try to connect to a cluster that is not running/responding, I do get the expected AeronTimeoutException from AeronCluster.Connect(). But additionally the ErrorHandler is called with a NullReferenceException.

The ErrorHandler is called from CloseHelper.Dispose:

at Adaptive.Agrona.CloseHelper.Dispose(ErrorHandler errorHandler, IDisposable disposable)
   at Adaptive.Cluster.Client.AeronCluster.AsyncConnect.Dispose()
   at Adaptive.Agrona.CloseHelper.QuietDispose(IDisposable disposable)
   at Adaptive.Cluster.Client.AeronCluster.Connect(Context ctx)

The NullReferenceException that is passed to the ErrorHandler has this StackTrace:
at Adaptive.Agrona.CloseHelper.Dispose(ErrorHandler errorHandler, IDisposable disposable)

The issue is that AsyncConnect.Dispose() calls CloseHelper.Dispose(errorHandler, ingressPublication) where ingressPublication might still be null. Additionally CloseHelper.Dispose(errorHandler, ingressPublication) is called twice in AsyncConnect.Dispose(), which looks unintentional, so I removed this too.