Solid angle per sky segment is hardcoded instead of read from the irradiance data

When converting summed radiances into an irradiance, we multiply by a fixed solid angle of 0.065 sr. That is only correct for HEALPix nside=4. At nside=8 the real value is about 0.0164 sr. The irradiance JSON already carries the right value in its metadata (nside and pixel_solid_angle). Neither field is part of the SolarIrradianceData type yet, so that needs extending too.

Another thing: a single factor here assumes all segments have equal solid angle, which holds for HEALPix but not for other sky discretizations like Tregenza. But I guess we say that HEALPix is needed for now.