Add Github style codeblock fencing

[gawhyte]

https://help.github.com/articles/github-flavored-markdown/#fenced-code-blocks

I think this would be a handy feature, but I understand if you don't want to poison the implementation with Markdown dialects.

I modified sub _DoCodeBlocks to do two passes. First it tries to find any fenced code blocks using code adapted from sub _DoCodeSpans, hashes them, then proceeds to perform the traditional codeblock check. I did not write any new tests, but it continues to pass existing ones.

[bobtfish]

I'm down with accepting this, but can you add a new test that tests this syntax?

[rlauer6]

btw - this PR fixes another issue - it would be great if it could be merged in and CPAN updated...

Here's some markdown that demonstrates the bug...

1. Line 1
1. Line 2
1. Line 3 with code block

   ```
   line 1
   line 2
   ```
1. Line 4 with code block

   ```
   a line

   a line after a blank line
   ```
1. Line 5

The bug renders the blocks as 'code' and fails to detect the block on line 4:

!

...with patch...

[h3xx]

Mostly works, but it appears to be double-processing code inside the fences. I.e. # foo --> <h1>foo</h1> --> <h1>comment</h1>.

Generates strange output given the following markdown:

```
# comment
```

Output:

<pre><code>&lt;h1&gt;comment&lt;/h1&gt;
</code></pre>

Expected:

<pre><code># comment
</code></pre>