Cleanup UTF-8 functions

[dfabulich]

From #998:

• Rip out encodeURI8859 and replace it with encodeURIComponent (this causes a ton of conflicts with open PRs, so I'll hold off for now, but it's a simple regex find-and-replace)
• Rip out rss_encode (ditto; lots of conflicts)

[salty-horse]

My understanding is that approx_utf8() is now only used for "normalizing" tags into a common spelling. The name of the function, and comments around its use are misleading.