Optimize and fix autolink function (#1442)

[mjwwit]

This PR fixes the markdown rendering issues with code-blocks containing URLs (#1442). The culprit was the autolink function, which conflicted with highlight.js. I've excluded code-block URLs from being auto-linked to fix this issue.

Besides being broken the autolink function was also crazy inefficient. I've improved the efficiency by making use of the browser-native createTreeWalker function. If this function is not available (IE8 and below) I make use of a recursive function (which is still more efficient than the previous implementation).

[strk]

Can you try to have this bug covered by an automated testcase ?

[mjwwit]

I'd love to, but there are no automated tests for the front-end AFAIK. Would you like me to build a unit testing environment just to test this case?

[strk]

I would *love* you to build a unit testing environment, not just to test this case :)

[andreynering]

Since we don't have a test setup for the font-end yet, this should not be a block for this PR. That should be discussed in a proposal first, and than made in another PR.

[strk]

I just asked for the test because a fix and a speed improvements are reported but wouldn't be easy to perform the test required to prove it. As we need 2 reviews, it'll take 2 people to perform independent tests. Having an automated test would just take verifying the test once :)

[mjwwit]

Reviewing the fix is quite easy. For your convenience I made a benchmark for the performance difference:
https://codepen.io/mjwwit/pen/YZbJEN?editors=1010

For those who can't be bothered to try it out, my optimized implementation will be about 30-50% faster on average.

[strk]

Nice app, codepen.io, first time I see it ! On Firefox version 45.3.0 on a 64bit GNU/Linux this is what I get for 1000 loops: JQuery: 451.3350000000082 ms TreeWalker: 474.62999999999556 ms Does it mean your change would make it slower there ? I guess Chrome based browsers would have much different numbers

[strk]

Same system, 10000 iterations: JQuery: 3595.914999999368 ms TreeWalker: 2506.635000000824 ms LGTM :)

[mjwwit]

Yeah, it can be quirky at times. I added a warm-up before the tests, but the results are still not super consistent. I guess it's because there are a ton of factors involved. On average though it will perform significantly better.

[lunny]

LGTM