HTMLSanitizer strips data-* attributes

[cdent]

Since the whitelist process is list-of-strings oriented, it will reject custom data-* attributes that are being used.

The hack to change it is described here:

https://github.com/rubys/venus/pull/19/files#diff-dcd60d82836175427dfbdf792aa976ecL172

I'm fairly certain that data-* is "safe" so it seems like there should be some way to keep it.

BTW: I looked and looked to see if this issue was already discussed somewhere (given the above link, I figured it would be) but I couldn't find it.