[Discussion] textContent vs. innerText

In #1764 we began a discussion as to whether Prism.highlightElement should use textContent or innerText to get the text of the element to highlight.

Summary:

• If the element to be highlighted doesn't contain any tags, innerText and textContent will return the same text.
• innerText only returns the human-readable text of the element which makes it more intuitive. textContent ignores the meaning of all tags. This causes linebreaks that are implemented via <br> or a wrapper with display: block to be ignored.
• The result of innerText depends on whether the element is visible. (If it's not visible it will return the same as textContent.)
• textContent is generally faster.
  innerText takes the current style of the element (plus children) into account (which might require the layout to be recomputed).

Additional concerns:

• Changing to innerText might break the code of some people.
• There are behavioral differences for innerText in some older browsers.
• We have to change the implementation of Keep markup, Line numbers, and Normalize whitespace when using innerText.

Not concerns:

• Both innerText and textContent have to good cross-browser compatibility with some older version of Firefox not supporting innerText and some older versions of IE not supporting textContent.
• We don't have to change the implementation of Unescaped markup when using innerText.

/cc @SetTrend @mAAdhaTTah @Golmote @LeaVerou

Start with the Prism.highlightElement entry point and review the Keep markup, Line numbers, and Normalize whitespace plugins named in the discussion. Compare the current behavior of textContent and innerText, including visibility and browser differences. Done requires an agreed choice and coordinated behavior across the affected implementation and plugins.