Incorrect parsing of code span

[dbuenzli]

> printf '\```ext``' | cmark             
<p>`<code>ext</code></p>
> printf '\```ext``' | cmarkit html   
<p>```ext``</p>

[dbuenzli]

The escaped backtick logic is wrong, somehow Cmarkit takes an initial escape in a backtick run as disabling the whole run.

We should either change the tokenization strategy or match against count - 1 when the run starts with an escape and textify the escaped backtick. Unclear in my mind right now but we need to be careful about this.