Forcing a tool to do something it was not designed to do adds complexity, maintenance overhead, and is bad engineering.
Calling a Common Lisp function from your Rail project might not be the best solution to your problem all things considered even if the Ruby version of that function is both slower and less elegant than the CL version.
It doesn't come down to it: it was about economics all along. All these philosophies -- polyglot, more monoglot, etc. -- have costs. You can choose to ignore it sometimes by staunchly recommending one approach over another in generic settings. But the underlying costs are still there.
Anyone who advocates pure monoglot or always using the best tool is not being pragmatic. Decisions should be driven by cost/benefit and not religion.