Integrate new JRuby reference pool

This work adds a new JRuby pool architecture that maintains a single a JRuby instance through which requests to Puppet Server will be run concurrently. This behavior can be toggled by setting the `jruby-puppet.multithreaded` setting to `true`. In this mode, the server's memory footprint will be significantly lighter due to no longer needing to run multiple JRuby instances. Please note that the work to ensure that Puppet's Ruby code functions correctly in this environment is still underway. This mode is considered highly experimental at the moment.