Question 1:
On that version of WMQ, no. You are much better off to back up the objects, security rules, exits and so forth and redefine the QMgr. Even on newer versions of WMQ where there are no issues with shared memory segment mapping, you are still much better off making a new QMgr by redefining the objects than by copying the run-time files over. This is because the new QMgr will end up with a new QMID and can therefore participate in the same cluster. (That said, having multiple QMgrs with the same name is an anti-pattern as I'll explain further below.)
Question 2:
No. A cluster is more than just a way to offload channel admin to MQ. It is a big namespace and within that namespace names are unique within a QMgr. IBM came up with the QMID that differentiates between two QMgrs as a way to distinguish when one QMgr had been deleted and rebuilt/replaced in a cluster. It was never intended to facilitate using multiple same-named QMgrs in the same namespace, though. To illustrate, imagine receiving a request message from one of the QMgrs. To which one does the reply go?