γ_t around GAIA stars and λ statistics

Supersedes #177. Martin's Feb-2026 work computes γ_t around GAIA stars (split by G magnitude into bright / medium / faint) and the λ statistics (plot_gammat_stars_around_galaxies, plot_dgammat_stars_around_galaxies, plot_dsize_stars_around_galaxies) from a CosmologyValidation that lives outside the repo (~mkilbing/.../rho_tau_stats; papers/catalog/run_cosmo_val_lambda.py imports it via sys.path). The in-repo cosmo_val has none of it, and its psf: block is consumed only by rho/tau stats (psf_systematics.py), so the GAIA config entries can't land alone.

Inputs already on candide: /n17data/UNIONS/GAIA/gaia_stars_g_{bright,medium,faint}.fits (Nov 2025).

Done means: gammat_around_stars / λ statistics implemented in src/sp_validation/cosmo_val (optionally tomographic), papers/catalog/run_cosmo_val_lambda.py importing from the package instead of a home directory, and cosmo_val/cat_config.yaml carrying the three GAIA versions from #177 (SP_v1.4.6 base, psf.path → the GAIA file, hdu: 1, square_size: True).

Start by comparing the in-repo src/sp_validation/cosmo_val with the external CosmologyValidation used by papers/catalog/run_cosmo_val_lambda.py, and review cosmo_val/cat_config.yaml plus psf_systematics.py. Done means the GAIA star and λ statistics are available from the package, the script imports the package, and the three GAIA configuration variants from #177 are present.