Allow multiple concurrent instances of benchmark-runner

[RobertKrawitz]

It is presently not possible to run two instances of benchmark-runner concurrently. While it would not be appropriate to do that when running benchmarks, for development or testing purposes it should be possible. Example use cases:

1. Run CI pipeline on multiple PRs concurrently on one cluster.
2. Allow functional testing concurrently with functional or PR CI runs on one cluster.

Currently there are two reasons why that can't be done:

1. The benchmark-operator is cleaned up with each run, and it currently always resides in one namespace (benchmark-operator).
2. The database pods for hammerdb run in fixed namespaces.

The second problem should not be difficult to solve, as we control those namespaces. The first one requires one of two things:

1. Whether it's possible to run two concurrent instances of the benchmark-operator in different namespaces.
2. Whether it's possible (or safe) to run the benchmark-operator without cleaning up.