Recursively skip verbatim elements

[mre]

This handles cases like

<code>
    <pre>
        <span>https://example.org</span>
    </pre>
</code>

where a link was embedded in a verbatim block, but embedded in a non-verbatim element (<span>).
This was correctly handled in html5ever, but not html5gum (which is our default HTML parser in lychee)

Fixes #259 (comment).

@untitaker, is there a better way to handle nested tags in html5gum or would you have resolved the issue in a similar way?

[untitaker]

@mre sorry for the slow response. I believe a better fix would be untitaker/hyperlink@07a7f20, which uses naive_next_state to make the parser not emit <a> tags at all inside <script> tags (and I think inside pre too)

[mre]

Oh I see. Would you have the time to create a pull request to change it to the new structure? As the author of the library it's probably way easier to do for you than for me. 😅 Maybe you want to ignore what I did in this PR and base it off of the last version of master just before this commit?

[untitaker]

Ah I see, you are extracting from text as well even if there's no <a> tag. I don't think it will help here. This is probably fine.

[mre]

Probably not the cleanest implementation, but yeah. Thanks for the feedback. 👍