I don't see it for 7-38; the three top level FCF callouts share a single DRF. Because of that they have a simultaneous requirement because they all have the same references. The "null datum" only has an effect when no datum references are made so they have the exact same simultaneous requirement since they share a single, null, DRF.
There is a special rule for composite FCFs so that they could not be applied individually meaningfully; the notation "SIM REQT" is there, but composite on a single hole?
If one wanted to, one could have two levels of individual segments applied to each hole to duplicate the one composite tolerance, identifying the holes in each group and adding a note , but that is complicated and composite is a more compact form. This would apply to this particular example as figure 7-38 does not duplicate any of the datum features referenced for location of the holes, which in composite tolerancing, are, by rule, deprecated to providing orientation only. However, with Voelcker's notation, that location capacity could be eliminated for the more general case.
I have no explanation for the often sloppy work done in the illustrations. I wrote before, if this was the work of a single author focused on the problem, they would need a mental health check up for dementia and to fire their editor. As it is, it's just incompletely reviewed and contains some fan fiction.
The same result as INDIVIDUALLY could have achieved the same by moving the callout to a third level FCF, as in 7-45. If there is a fight between INDIVIDUALLY and SEP REQT, which look to me like synonyms, it's independent of not having datum references. The choice of which to use is decorative, not definitive.
I could place a hole dimension and size and location tolerance table on the drawing and mark an identifier next to each hole the values applied to and exactly define the members of the group and never put down #X_ and that should be acceptable.