Adopt r_p (physical separation) for clustering/GGL; keep shear angular #276

Do galaxy-galaxy lensing and clustering in transverse physical separation r_p (Mpc/h) rather than angular theta where spectroscopic z is available: cleanly isolates scale ranges and gets the BNT scale-cleaning benefit for free. Cosmic shear stays angular. Delta-Sigma in physical units; number of radial bins (6-7) TBD. Ties into the UNIONS-WL/Smokescreen fork PRs (blinding, modelling arbitrary data vectors).

Context: Paris meeting notes