In that case you shall not add this .002 to profile tolerance value, otherwise you may end up with accepting a part that does not meet profile requirement.
You may think of it in following way [let's say the basic rectangle (defined by basic dimensions) is 5" long and 2.5" wide and the datum feature hole is 1" +.002/-.000)]:
Imagine a wireframe containing following elemets:
1. Circle of dia. 1" simulating datum feature hole at MMB
2. Two rectangles - 5.05"x2.55" and 4.95"x2.45" simulating profile tolerance zone. The rectangles are perfectly centered to each other and to the circle defined in point 1.
Now, take your part and try to fit it - by rotating and sliding - inside the wireframe, so that the surface of datum feature hole does not violate circle from point 1 and the outline of the rectangle fully fits inside a space between two rectangles from point 2. If you weren't able to fit it, the part is out-of-spec.
So what is the benefit of (M) modifier in profile FCF? It is hidden behind a type of movement you did when trying to fit the part inside the wireframe. If you were able to slide and rotate the part that means the diameter of datum feature hole was bigger than 1.000 and smaller than 1.002. If you were only able to rotate the part, the diameter of datum feature hole was exactly 1.000. Rotation would be also only type of movement available if the profile FCF was [.005].
Having said all of that, notice that (M) modifier does not enlarge profile tolerance zone - it is always 0.005 wide. Instead of it you gain a possibiliy of shifting the part to fit it inside profile tolerance zone, if datum feature hole is not at its MMB = 1.000.
I hope you picture what I tried to say.