[Feature request] Ability to specify number and location of dqlite voters

While MicroCeph can scale beyond three nodes, once it does, there only remains three dqlite voters. These voters are assigned automatically and the quantity and location of the voters cannot be changed.

We had an instance during testing where 2/3 dqlite voters in a five-node MicroCeph cluster ended up in the same failure domain (rack). When both of these dqlite voters went down, dqlite/MicroCeph lost quorum while Ceph quorum would have been fine with 3/5 members.

It would be nice to be able to increase the number of voters and/or specify the location. Right now, some creative rebooting of the cluster members have allowed us to make sure that all three voters are in separate failure domains/racks.