PREOPS-4949: Add stackers useful for schedview (Teff and dayobs related)

[ehneilsen]

I'll do the addition of the plotters we discussed as a separate PR.

[rhiannonlynne]

I realized I had a concern that if there were two bundles with TeffStacker but one "normed" and one not, that they might run at the same time and have a conflict about the t_eff column. However, apparently we built MAF better than that and it does the right thing (runs them separately, rewrites the t_eff values, and comes up with the correct answers. Yay!

import rubin_sim.maf as maf

opsdb = '/Users/lynnej/opsim/baseline_v3.4_10yrs.db'

teff_stacker = maf.TeffStacker()
teff_stacker2 = maf.TeffStacker(normed=True)

bundle = maf.MetricBundle(maf.SumMetric(col='t_eff'), maf.UniSlicer(), '', stacker_list=[teff_stacker])
bundle2 = maf.MetricBundle(maf.MeanMetric(col='t_eff'), maf.UniSlicer(), '', stacker_list=[teff_stacker2])

g = maf.MetricBundleGroup({'teff': bundle, 'teff_normed': bundle2}, opsdb, verbose=True)
g.run_all()

bundle.metric_values, bundle2.metric_values

produces

(masked_array(data=[44791395.476532385],
              mask=[False],
        fill_value=-666.0),
 masked_array(data=[0.6969573084883421],
              mask=[False],
        fill_value=-666.0))

as expected.

values from the previous TeffMetric for the same run:
All sky all bands 0.70 (normed) 44791395.48 (total).

[rhiannonlynne]

You might as well export "compute_teff" as well, though maybe? (especially if exporting the fiducial values).
In theory, you could use the compute_teff with a pandas data frame using apply, although it may be better overall (i.e. from schedview) to just use the
maf.get_sim_data() method, pointing to the visit database in question, and add the stackers you want.

[rhiannonlynne]

Ooh, right -- when using Normed stackers, the sum becomes a MeanMetric.