Zipped html import is only partially fixed, I reopened that bug. I'm only seeing html files that are part of an uncompressed epub being rezipped.
Also seeing another new bug - regression from beta 6 - where I can't convert from zipped html to epub because beta 7 seems to be extraordinarily picky about which css fields and values are acceptable. What makes it particularly odd is I'm working with css files that were generated by beta 6/7 Lit->epub conversion.
Bug references:
zipped html:
http://calibre.kovidgoyal.net/ticket/2666
css errors:
http://calibre.kovidgoyal.net/ticket/2628