My pdfs with problems for CE characters are pdfs made directly from OCR by someone, no editing or correcting, they are old and have some unusual fonts embedded by the OCR software, custom encoding, so it's no wonder they can't show properly in duokan. What I mean is that's not a duokan bug. Kindle firmware also have trouble with them, showing spaces instead those CE characters. Although those pdfs were properly shown in older versions of duokan.
Every other pdfs are ok in duokan.
1. Maybe you could remove the hyphenation files, see what happens
2. There's no fixed page layout in mobi, as in pdf or paper book, so you can't number pages, cause pages do not exists. You confuse page with screen area. If you change font type/size, you make different "pages" with more/fewer words, so there's no point in page number, only percentage based on total characters number, cause that's permanent.
3. Bookmarks are based on percentage, not on "pages". Next page is actual next characters till the end of screen or end of chapter