Stop errorfactory from creating errors with nil causes

[jgibson]

Q	A
Bug fix?	yes
New feature?	no
API breaks?	no
Deprecations?	no
Related tickets
License	Apache 2.0

What's in this PR?

Currently when the errorfactory is invoked with a nil error the errorfactory will produce an error that has nil error field. This PR will create a new underlying error when presented with a nil error.

Why?

Some consumers of the error assume that the error has been populated and when they access the field it triggers a panic. In particular logging code like this in nificluster_controller.go fails:

r.Log.Info("Nodes unreachable, may still be starting up", zap.String("reason", err.Error()))

Some error producers need to report errors even the underlying failure doesn't produce an error itself. For example in client.go:

clusterEntity, err := n.DescribeCluster()
if err != nil || clusterEntity == nil || clusterEntity.Cluster == nil {
	err = errorfactory.New(errorfactory.NodesUnreachable{}, err, fmt.Sprintf("could not connect to nifi nodes: %s", n.opts.NifiURI))
	return err
}

Additional Context

This failure made it more difficult to diagnose some underlying connectivity and configuration issues that I encountered when I moved on from an insecure cluster to an SSL-secured one.

Checklist

• Implementation tested
• Append changelog with changes

[juldrixx]

Can you please add your change in the CHANGELOG?

[jgibson]

OK, I've updated the CHANGELOG. It seemed presumptuous to update the CHANGELOG before a maintainer accepted it.