don't read two copies of the request body in for JRuby requests

Due to some conflicting opinions about how request bodies should be processed between the Ruby Puppet code and ring's 'params' middleware, our current code has the unfortunate behavior of reading the body of the HTTP request into memory twice for any requests that are routed into the JRuby layer.

We can prevent this by duplicating a few functions from the ring middleware library in our own code base, which isn't awesome, but seems preferable to the unnecessary extra memory usage (which can be a pretty big chunk of RAM for requests like catalog/report).
_________
Risk Assessment: Low
Severity: Low
Probability: Low