I think search head pool is required for multiple search heads sharing configurations.
What's the benefit of making it visible from search peers?
The other benefit is just the increased efficiency obtained by not having to replicate knowledge bundles across the set of indexers with each new search. For more information, see:
http://www.splunk.com/base/Documentation/latest/Deploy/Mounttheknowledgebundle
The benefit of sharing the configuration between your search peers is that you won't have to replicate bundles all the time. While replication is occurring, searching can't happen. So your going to mitigate potential issues with search by sharing configurations.
4.2 does allow asynchronous replication during which searches can run, but there are certain effects this can have particularly if you have large bundles.