Enhance NLP to understand # Questions

[danielreed]

When we support statements like

@oneview 3 questions. What server profile template is SY-BL-234562 using? How many servers are using that template? What is the average CPU utilization of those servers?

We need to be able to normalize a statement like the following to inject a # questions quantifier so we hit the appropriate regex in hubot.
What server profile template is SY-BL-234562 using? How many servers are using that template? What is the average CPU utilization of those servers?

The NLP framework already identifies sentence structure, we could just inject a # sentences construct to the front.