(a) The electric field due to a discrete point charge is not well-defined exactly at the location of the charge — as you approach a point charge, the field's magnitude grows without bound (E∝1/r2→∞ as r→0). A Gaussian surface passing through this point would make the field undefined on part of the surface, so flux cannot be meaningfully calculated there.
(b) Yes — a Gaussian surface can pass through a continuous charge distribution (like a uniformly charged solid sphere), because the volume charge density there is finite, so the electric field remains finite and well-defined at every point, including inside the distribution itself. (A thin charged sheet is a special case: the field jumps by σ/ε0 across it, but a pillbox meets the sheet only along a curve of zero area, so the flux is still well-defined.)