Underestimated F090W fluxes and improper F090W kernels.

[gnoir0t]

Hi Gabe,

Reporting the F090W issue here so you can have a look and hopefully figure it out.
When extracting and fitting data in the Technicolor fields with grizli, we noticed a systematic flux drop in F090W, creating artificial breaks especially noticeable for brighter sources (see attached image).
We noticed that the F090W inverse sensitivity curves for both grisms were off, and we rescaled them (for the first order only, see attached plots).
This rescaling of the inverse sensitivity curves (NIRISS.GR150C.F090W.p1.etc.sens.fits and NIRISS.GR150R.F090W.p1.etc.sens.fits) made only a little improvement and the flux drop/underestimation largely remains (see plot).

Also, this might be unrelated, but the F090W kernels seem screwed up (see plot).