Backup bucket endpoint errors if cluster is in mixed-mode due to _system scope

What is the problem?
When a cluster is in mixed-mode with at least one pre-7.6 node and the query node is 7.6.x then hitting a bucket's backup endpoint fails with the following error message:

❯ curl 10.145.246.101:8093/api/v1/bucket/default/backup -uAdministrator:asdasd

{"_level":"exception","caller":"couchbase:2994","cause":{"scope":"default:default._system"},"code":12021,"key":"datastore.couchbase.scope_not_found","message":"Scope not found in CB datastore default:default._system"}⏎

It looks like the reason is the bucket endpoint handler assumes that the _system scope is created and so scans it for metadata objects, but this is not true in this case - I believe the _system scope is only created once the upgrade has complete.

Reproduction

1. Create a cluster with a 7.2.4 node and a 7.6.1 node, with the latter having query
2. Create a bucket
3. Try to backup the bucket using cbbackupmgr