Every cluster member sends the same index updates to a remote Solr server

Update of the proposed fix: it now adds isIndexShared() to the internal SolrConfiguration role (true when solr.type is remote), and both listeners skip a remote event when that method returns true. This replaces the check on the type mentioned in the previous comment; the behaviour is the same. XWIKI-25254 and XWIKI-25259 use the same method, so the assumption "a remote Solr server is shared by all the cluster members" is defined in one place. No new configuration property: whether isIndexShared() should be overridable by one is left to the review.