Cosmology forecasting prerequisites

[aimalz]

We need more information about the simulations to get a Fisher forecast on cosmology from n(z)s derived from samples with and without CASTOR data, specifically a covariance matrix (per tomographic bin). Then we can run it through CHIPPR's CosmoLike procedure (or a modern analogue -- strongly preferable because CHIPPR is still in Python 2, sorry).

[aimalz]

Update: recalling more about how the simulations were made (from COSMOS narrow-band as truth), we need to find a way to fudge covariances (with LCDM approximations or autodiff/jaxcosmo?) or use other simulations (difficulty unclear).

EDIT: better idea: Use the metrics from the LSST-DESC Tomographic Binning Challenge!