commonmark.js adds a language- prefix to the info string provided after the opening code fence, regardless if it already starts with language-.
Could commonmark.js not adding the language- prefix in this case to allow using the language- prefix in the source whitout resulting in language-language-infostring in html?
E.g., using the following source:
```language-r
x <- 1
```
```r
x <- 1
```
We get the following result:
<pre><code class="language-language-r">x <- 1
</code></pre>
<pre><code class="language-r">x <- 1
</code></pre>
The commonmark spec currently says that it "does not mandate any particular treatment of the info string". A language- prefix is added in example 142.
With pandoc --from commonmark --to html5 --no-highlight, we get:
<pre class="language-r"><code>x <- 1</code></pre>
<pre class="r"><code>x <- 1</code></pre>