Create separate method for rendering exercise HTML

[gadenbuie]

I'm moving this to its own issue because it's a larger task than where I originally proposed it. In short, we could reduce complexity and improve extensibility and maintainability of the exercise components by

1. Using the knitr hooks to create the exercise object and set required knitr chunk options
2. Using a separate function to format the exercise component into its initial HTML format

Originally posted by @gadenbuie in #548 (comment)

I think the solution here is to remove the complexity around the knitr hooks. Instead of individually writing the tutorial support chunks, we should have the exercise chunk write all of the HTML for the exercise and then we'll disable any kind of output from the support chunks. In addition to simplifying the logic, this would let us re-use chunks between exercises.

Alternatively, if we decouple hints and solutions and let the shiny app control whether the solution button is present (and what that solution is), then we wouldn't even need to write the solution chunk html into the document.