`_` treated as `<em>` when it shouldn't

[justanotheranonymoususer]

Version: 8c74601

import markdown2

# Expected output:
# <p>_word1 <strong>strong</strong> word2._word3</p>
# Actual output:
# <p><em>word1 <strong>strong</strong> word2.</em>word3</p>
markdown_text = R"""
_word1 **strong** word2._word3
"""
print(markdown2.markdown(markdown_text))

[twavv]

This also happens with URLs which breaks them.

import markdown2
print(markdown2.markdown("https://docs.google.com/document/d/id_with_underscores/edit"))
# <p>https://docs.google.com/document/d/id<em>with</em>underscores/edit</p>

[Crozzers]

This also happens with URLs which breaks them.

Fixed with middle word em extra.

---

Regarding the main issue, according to the original markdown spec this is a valid em.

However, the original spec is vague and this example does look incorrect. The GFM spec covers this and has a comprehensive set of rules attached, but I don't see an easy way of implementing that.
It would likely require completely re-writing how we process emphasis, doing away with the regex and going for a manual approach