Fix pre issues

[jnvsor]

Fixes #70

Far slower performance with blanklines enabled though.

dab26c8 blanklines is correct as far as browser output goes (Extra whitespace is only in markup, not rendered) and is a whopping twice as fast as this version. (And faster than current master too) so perhaps it would be a better idea to just reset to that and ignore the whitespace inconsistancies.

Also, this shines light on the fact we need a second benchmark for blanklines mode - I'll get to work on that now

[jnvsor]

That's weird - the test passes on my machine.

At this point @penibelst it's becoming a right pain in the ass and I think we should go back to dab26c8 on the whitespace

[jnvsor]

Closing this, as discussed in #70 I don't think the performance penalty is worth it. I'm going to revert to a simpler version and add another PR