replicate bibliographic solr

[solrize]

Set up a new solr on ia331507, start a new update daemon to keep it refreshed, and patch the search plugin to split requests across multiple solrs, failing over automatically when one or more is down.

Desirable: concoct some scheme to make multiple solrs less likely to do expensive merge operations (degrading responsiveness) simultaneously. Simplest approach is probably to make the commit intervals unequal or random.

[solrize (solrize)]

Also desirable: have a cron job that optimizes each solr once a day (not simultaneously). It should also save a snapshot of the index, so we can roll back to something fairly recent in case some update causes corruption.