Events and notifications stay disabled until restart when the remote Solr server cannot be reached during startup

Update of the proposed change: the lazy resolution of the events core is now shared with XWIKI-25257. Instead of its own copy of that logic, the Solr event store uses the same internal helper as the search core and the extension index (the core is requested again when needed, not more than once every 10 seconds, with one warning when it is not available and one information message when it is available again), so this change now depends on XWIKI-25257.

As a consequence, an embedded events core is checked strictly at startup again: with the embedded Solr, an events core which cannot be loaded fails the initialization of the event store, as before this change and like the search core, since it is a configuration error rather than a temporary condition. The tolerant startup only applies with a remote Solr server (solr.type=remote), which is the case this issue is about. The change of RemoteSolr (the initialization no longer fails when the server cannot be reached) is unchanged.

The log messages are now the helper's: "The Solr core [events] is not available: [...]. Check that the Solr server can be reached and that the core exists on it. The core will be requested again when needed." and "The Solr core [events] is available again".