As I explained in my reply to Mr Everson last month, I'm not sure that any more than an extension of 5 letters would be required, if unification with Shavian were to occur:
The reason being, every existing Shavian codepoint directly corresponds with a QS letter (or sequence of two letters). If we are to go with unification, why add new encodings for letters that are direct, 1:1 (or many:1) equivalents? (even if they switch around shapes or replace them entirely.)
By all means, see my reply to his post "QuikScript and Unicode" for the full argument.
And just to play devil's advocate regarding your suggestion of 16 characters: /eɪ/ and /aɪ/ may have different shapes, but only barely. The top bar of Shavian 𐑱/𐑲 was switched out for a little loop at the top. I've never seen this an impediment to reading. Further, the only difference with /b/ is that it was mirrored horizontally in QS, with a loop added so that it ends on the baseline instead of the descender line.Those are pretty minimal changes in my opinion.
What do you think?