Encode supplementary characters in enriched link titles

[rizen]

Summary

• encode four-byte characters from remote page titles as numeric HTML entities
• preserve the visible character while keeping generated HTML compatible with MySQL three-byte UTF-8 fields
• prevent database truncation from leaving the rest of a shop page inside an open link

TDD evidence

• Red: author.t/contentfilter_supplementary_title.t returned the raw four-byte balloon character
• Green: prove -lv author.t/contentfilter_supplementary_title.t author.t/markdown.t passes 24 assertions

Fixes plainblack/TGC#1549