Hi Nate – would you mind performing a quick review of this?

This is a port of HSC-1005. Specifically, we are adding a new slot, CalibFlux, which we use to record the flux measurement to be used for calibration. There are changes in six repositories:

• On afw's tickets/DM-3106 we define the new slot and add some tests to ensure that it (and other flux slots) successfully provide basic functionality;
• On meas_base's tickets/DM-3106 we arrange for measurement tasks to store an appropriate value in the new slot;
• On tickets/DM-3106 in ip_diffim , meas_algorithms and meas_modelfit we simply tweak some task configurations to avoid recording a CalibFlux when it's not necessary;
• On pipe_tasks` tickets/DM-3106 we set the photoCal task to use the CalibFlux slot rather than the PSF flux measurement.

Note that HSC-1005 includes a few other tweaks to obs_subaru related to using an aperture flux for photometric calibration. We're not currently doing that on LSST, so they don't apply here. We'll consider how to bring them over as part of DM-3108.

Nate – thanks for the review! A brief response:

rename baseName to base, to be more consistent with how slots and such a documented in other places

On reflection, I agree that using "baseName" is a poor choice. However, I also don't like "base" – it's normally used to indicate the output of a measurement plugin defined in meas_base (cf the prefix "ext" for meas_extensions_*). Instead, I will change it to use "afw_Test", since this is a test in the afw package.

consider using python .format() for your string formatting. It will help future proof things, and be a bit easier to read.

You may consider it considered!

For what it's worth, I find it hard to get really excited one way or the other. When this was discussed on HipChat a few months ago (in the Tavern, so carrying no formal weight), there was no real consensus about which was the better approach; while a couple of folks are strongly in favour of one approach or the other (no prizes for guessing names), I think the general attitude was that we expect both to be supported in Python for the long term and that which is more readable depends chiefly on context.

Here, % formatting is used extensively in the other afw tests, while I don't think any of them call str.format(); for consistency, I think we should stick with the former.