Cannot compute cross-spectra from two sets of alms with same fields

[ntessore]

Computing the cross-power spectra from two sets of alms with the same field names has unexpected results:

alms_lensmc = heracles.read_alms(...)
alms_metacal = heracles.read_alms(...)

# does the wrong thing if both alms have fields called "SHE"
cross_cls = heracles.angular_power_spectra(alms_lensmc, alms_metacal)

Here cross_cls is computed as if it was an auto-correlation, including bias subtraction and treating SHE, SHE, i, j as equivalent to SHE, SHE, j, i.