What do you think of my example of the pakistani tanween with small
meem, indicating tanween + iqlaab, which from the grapheme point of
view is in addition to and offset from the tanween?
(http://kprayertime.sourceforge.net/calligraphy/tanween-dammataan-iqlaab.png
)
Doesn't this indicate that iqlaab should be encoded as such, and not
incorporated into the tanween?
Well, in my view this is an example of how not to identify graphemes. The
Egyptian and Saudi editions express iqlaab with a ligature of vowel and
small meem, your example shows a tanween ligature with small meem, but the
underlying grapheme is identical: tanween+iqlaab.
The first thing to agree on is to encode iqlaab as a separate grapheme. What
rests then is how to encode tanween. Unicode adopted the tanween ligatures
as separate codes. My opinion is that the ligatures fathatan, dhammatan and
kasratan are not graphemes, but ligatures consisting of exactly what their
Arabic names indicate: two fathas, two dhammas and two kasras.