Regression for **bolded_code_friendly** text

[mhils]

Describe the bug
Starting with fcaadfe, markdown2 renders _ as <em> in bolded text, even if code-friendly is set. @Crozzers, any idea on what's the best way to fix this?

To Reproduce

**bold_but_not_emphasized**

Full test case: mhils@bf39e38

Expected behavior

_ should stay as-is.

Additional context

Thank you so much for the other changes in 2.5.0, it's fantastic! 🍰

[Crozzers]

When I added the CodeFriendly extra class I had to change the program flow a bit, since it was no longer directly part of the _do_italics_and_bold function.

Previously we had two different versions of the strong/em regexes and we'd run one or the other. That's not possible in the new Extras format so the CodeFriendly extra runs before italics and bold and hashes text to protect it from the regex.

Currently it only hashes text if it's using the _underscore bold syntax_, but for *star syntax* it doesn't check if the text contains any underscores, which is how this can slip through.

I'll modify the extra to check for this and hash the text to protect it