Healpix <-> Rectangular needs to ignore nans

[greglucas]

Currently, it sounds like if we get a nan value in the healpix subdivision, we are "poisoning" the entire rectangular grid cell with that result. We should ignore those nan values in our calculations, and only set the value to nan if all healpix values are nan within that rectangular cell. i.e. if we have one good healpix value and 5 nan's, we should only use that one good value with the associated exposure time / angle etc. and not anything else.

I think this is the relevant area of the code.

imap_processing/imap_processing/ena_maps/ena_maps.py

Lines 1672 to 1679 in 3fe8424

	# Weighted mean (weighted by solid angle) of these values over the pixel axis,
	# which is the last axis of this array
	weighted_hp_vals_at_rect_pix_ctrs = (
	hp_vals_at_rect_pix_ctrs * rect_subpix_solid_angle_by_lat
	)
	mean_pixel_value = (
	weighted_hp_vals_at_rect_pix_ctrs.sum(axis=-1) / full_rect_pixel_solid_angle
	)

Relevant tests here:

imap_processing/imap_processing/tests/ena_maps/test_ena_maps.py

Line 1453 in 3fe8424

def test_to_rectangular_skymap(

imap_processing/imap_processing/tests/ena_maps/test_ena_maps.py

Line 1306 in 3fe8424

def test_calculate_rect_pixel_value_from_healpix_map_n_subdivisions(

imap_processing/imap_processing/tests/ena_maps/test_ena_maps.py

Line 1397 in 3fe8424

def test_get_rect_pixel_value_recursive_subdivs(