Update PSF modeling configs appropriate for LATSS

Namely:
(a) set the characterizeImage kernelSize and stampSize to 71
(b) change finalize to use a kernelSize and stampSize of 71
(c) change finalize PSF model from piff to psfex

Note that (c) is required (for now) because piff doesn't scale well
with increasing stampSize.

config/latiss/characterizeImage.py

@@ -15,6 +15,9 @@
 
 # Reduce psfex spatialOrder to 1, this helps ensure success with low numbers of psf candidates.
 config.measurePsf.psfDeterminer["psfex"].spatialOrder = 1
+# Set the default kernel and stamp sizes for PSF modeling appropriate for LATISS
+config.measurePsf.makePsfCandidates.kernelSize = 71
+config.measurePsf.psfDeterminer["psfex"].stampSize = 71
 
 config.installSimplePsf.width = 21
 config.installSimplePsf.fwhm = 2.355*2  # LATISS plate scale is 2x LSST nominal

config/latiss/finalize_characterization.py

@@ -1,2 +1,10 @@
-# Reduce piff spatialOrder to 1, this helps ensure success with low numbers of psf candidates.
-config.psf_determiner['piff'].spatialOrder = 1
+# Change the PSF determiner to psfex from piff default as the latter will not
+# scale well to the large kernel/stamp sizes for PSF modeling for LATISS data.
+import lsst.meas.extensions.psfex.psfexPsfDeterminer
+
+config.psf_determiner.name = "psfex"
+# Reduce spatialOrder to 1, this helps ensure success with low numbers of psf candidates.
+config.psf_determiner["psfex"].spatialOrder = 1
+# Set the default kernel and stamp sizes for PSF modeling appropriate for LATISS.
+config.psf_determiner["psfex"].stampSize = 71
+config.make_psf_candidates.kernelSize = 71

ups/obs_lsst.table

@@ -9,6 +9,7 @@ setupRequired(astro_metadata_translator)
 setupRequired(obs_lsst_data)
 setupRequired(daf_butler)
 setupOptional(meas_extensions_gaap)
+setupOptional(meas_extensions_psfex)
 
 envPrepend(PYTHONPATH, ${PRODUCT_DIR}/python)
 envPrepend(PATH, ${PRODUCT_DIR}/bin)