Changing locality of processes

We managed to lose data on one of our test clusters by executing the following steps:

1. We had 5 machines running FDB 6.2.1 - running 12 procs each
1. Additionally we had one machine that ran an older version (the processes there couldn't join the cluster)
1. The cluster was configured with `configure new triple ssd` and afterwards we populated some generated test data (TPC-C)
1. We changed the config with `configure ssd three_data_hall` which caused the cluster to go down (this is expected as we didn't set the locality_data_hall)
1. We changed the config on all machines and set `locality_data_hall` so that two machines would always share the same value.
1. We restarted fdb on all machines through ansible.

After that, all data was gone.

This being said: I am not 100% convinced that this is the full story and I am not sure this was an fdb bug. However, it seems that changing localities on processes isn't super well tested in simulation. Because of that I think we should mention in the documentation that changing localities on processes is not supported.

No file or test is identified in the issue. Review the documentation covering `configure`, `locality_data_hall`, and process localities, then verify the supported behavior with maintainers or existing tests; done means clearly documenting whether changing process localities is supported and any required procedure or warning.

Written by the indexing model from the issue text.