I think the problem is in the write function of the PDF class. It 
supports justified text, but it is impossible to change the font in 
the text string.
Besides, the function sends the data directly to the output stream, so 
you can't use 2 write calls.

I don't have a nice solution (yet). My hack solves the problem for me, 
but probably not for many others.

This sounds really hackish. But without a patch, I'm not even sure I 
understand what you suggested.

Justified text looks really nice and the File_PDF class can do this 
with the multiCell function.
BUT if you want to mark a word bold/italic inside this text there is 
no simple way to get the whole line justified.

I wrote a dirty hack. (It's very, very dirty, so I won't post a patch 
(yet)) But the idea is the following:
1. Add some HTML/BBCode tags to the text.
2. While parsing the text char after char, use the tags to update the 
font internally (Don't print the new font to the buffer.) but ignore 
them in the string width.
3. When formating the text for output (done in cell()), replace the 
HTML tags with something like this: ") Tj <NEW FONT INFORMATION> ("
The PDF viewer does the rest.

It took me some hours to figure this out. But with some help, maybe I 
can write a nice patch.