In fact, I use the Windows version, so it is more 
difficult to suppress foreign codes, than on UNIX.

But keeping only html code (or not) is like the options 
you already propose, transfering or not jpegs, zips, or 
writing ISO9660 names, ...

Moreover, you propose an option for detecting links even 
in javascript codes. You just have to suppress everything 
between <script> and </script>.

Keeping only html code, is typically an option for an 
offline browser that could seriouly diminish the size of a 
given site.