When clothifying a body part, the vertices welded to the adjacent body parts are automatically added to the _constrained group, thus ensuring that the connection remains intact.

Indeed the cloth room could be used for hair. Transmapped hair is nothing else than a veil or a hood with a hair texture instead of a cloth texture. I never tried it, but I think you could make transmapped hair realistically be 'squeezed' between a hat and the head.

By the way: strand hair can be manipulated using magnets and other deformers. Yuo an use this to generate the effect of pins and bands. In the basis it is all about moving vertices around. Whether these are used for geometric definition of splines(hair), surfaces carrying textures (clothing, props, figures) or solids (not supported by Poser) should make no difference.

The problem, and the obfuscation, comes from the way this is all packed and bundled in manipulation tools. The application designers must strike a balance between flexibility (steel learning curve, high level abstraction thinking), and conveinience of use for most common tasks. I would love to see flexibility and transparancy at the level of for example the  PovRay scene definition language, in which vertices, normals, colors, translations, rotations, scales and if you want box objects have the same definition: a tupel of 3 float numbers. I realize however that working on that level you can not make a practical application for the general public (which Poser is) and do things like dynamic clothing.  One has to make choices. Those who ever were in a jetplane and saw right below their own home which they left 4 hours before in a hurry to get to the airport in time to catch that flight will understand. It is our own fault. We do not have a main airport in our backyard.