HSC Fourier 3x2pt: N(z) calibration

[damonge]

• [ ] We need a DIR-type calibration stage (minimally using kd trees and nearest neighbor search).
• [ ] The associated spectroscopic catalog should be taken in as input at the selector stages.

Associated issue in TXPipe here.

[fjaviersanchez]

This may live outside of TXPipe, e.g., see FlexZPipe.

[aimalz]

I second what @fjaviersanchez said above. The infrastructure doesn't exist yet, but PZ aims to have a pipeline for getting N(z) by multiple methods regardless, so we can propagate the impact of various systematics in the photometry. If the 3x2pt pipeline could be structured to make a call to an external module, it would then be easier for the 3x2pt pipeline to adjust when the generic version from PZ is ready. Also, I think being outside of TXPipe might mean we can directly use @AngusWright's existing R version as is immediately rather than writing something from scratch only to replace it with the generalized PZ version (hopefully soonish).