Select a HTML reading library

[houfu]

We should pick a library that supports reading HTML markup.

• BeautifulSoup is probably the best and safest choice

Then, state it as an optional dependency.

Write initial docs, including limitations.

KIVs:

• Highlighting formatting differences seem to be in reach but it might need adjustments to core code
• Other HTML-like formats like RSS and XML may be possible, but it's unclear if it needs to be separated into its own implementation or treated as HTML-like.