update markdown rendering approach

[miguelcobain]

What this PR does:

• updates marked and highlight.js to the latest versions
• changes the approach to markdown parsing

Previously we were parsing the markdown with the vanilla tokenizer from marked. This meant that we had to make some massaging of the tokens to successfully remove extra paragraphs when inside a curly or angle bracket syntax (see compactParagraphs function).

This new proposal makes use of marked's extensions, trying to accomplish a similar same thing.

One advantage is that the content inside an html or hbs block now is preserved, including whitespace.
As an example, previously compiling this markdown:

{{#foo-bar}}

    hello

{{/foo-bar}}

resulted in the following hbs:

<div class="docs-md"><p>{{#foo-bar}} hello {{/foo-bar}}</p></div>

But now it results in the the following:

<div class="docs-md">
<p>{{#foo-bar}}

    hello

{{/foo-bar}}</p>
</div>

Notice how everything inside {{#foo-bar}} and {{/foo-bar}} is preserved. Same thing happens for html/angle bracket components.
This explains why I had to change tests.

Test coverage was good, but I went manually to every single page in the docs app and everything looks good.

Since @dfreeman seems to have worked on this code in the past, I'm going to mention him. Would be great to get another pair of eyes, at least in the test changes. :)

[miguelcobain]

htmlbars is no longer a thing in highlight.js, being replaced with just handlebars.

[miguelcobain]

highlightBlock method was deprecated, being replaced with highlightElement.

[SergeAstapov]

@miguelcobain This is great! Thanks for working on this!

[RobbieTheWagner]

This looks amazing to me, thanks for doing this @miguelcobain!