CDs reject requests using different connection string

Right now, `Coordinators` in FDB can serve any requests from any cluster. In some cases, this can cause potential outages. For example CDs from cluster A can serve requests from cluster B, but if the FDB binary versions in cluster A and B no longer matches, there could be an outage. And this outage is currently hard to root cause without good understanding how processes interact with CDs.

To simply that, a coordinator should reject requests that come with a different connection string than its own. This means a coordinator can now only serve one cluster after the change, but luckily there seems to be no use case of the multi-cluster support from the beginning.