This question is inspired by Colin McFauls very reasonable classification of questions tagged as homework due to the very broad notion of the term "homework-like" currently applied at Physics SE, and can be considered as a follow up of John Rennie's meta post asking for graduate-level upward questions not getting closed as homework.
I basically agree with John Rennie that gruaduate-level upward (technical) questions should not be closed as homework and want to go even farther by proposing that, refering to Colin McFauls classification, category 3 (not yet a good question but salvagable) and even more so category 4 (high-level problems) should not even be tagged homework. So my question is:
Can the very broad notion of "homework-like" as currently applied, be narrowed down to not include technical advanced topic questions that come up from reading higher-level textbooks, following advanced courses, or even from reading research papers (graduate-level upward for short)? And can then these advanced technical questions (not all of them at once of course) be detagged from homework (after appropriate edits if needed for category 3 questions)?
Here are some examples of advanced technical questions that should not be considered homework and therefore either directly or after an improving edit, detagged. I am not trying to sharply distinguish between category 3 and category 4 of Colin McFaul's classification at the moment, and this list is by no means comprehensive but contains just a few examples to show what I am talking about:
Conserved topological charge for d=3 Yang-Mills. G=U(2)
Correlators in String Theory (Green Schwarz Witten)
A question for the generalization of gauge transformation with two antisymmetric indices
Simple argument for unexpected behavior in SUSY model
Einstein action as a functional of the tetrad (first order formulation of gravity)
Deriving Virasoro algebra question
few fermions in a harmonic trap — position density matrix from diagrammatics
"Redshifting" of forces in stationary spacetime
...etc