DM-32601: Move remaining jointcal tests to gen3

Readme.md

@@ -7,10 +7,8 @@ It was derived from [meas_simastrom](https://github.com/lsst-france/meas_simastr
 ## A note on numbers in tests
 
 All of the `tests/test_jointcal_<obs>.py` tests run jointcal on their respective `testdata_jointcal/<obs>` dataset.
-The numeric values being tested (the `metrics` dictionary, and the `pa1`/`relative_error`/`dist_absolute_rms` values) were not computed in advance, but are empirical, from the first time that test was successfully run.
+The numeric values being tested in each test's `metrics` dictionary were not computed in advance, but are empirical, from the first time that test was successfully run.
 The one exception is the `cfht_minimal` dataset, which was designed to be small enough (3 `MeasuredStar`, 2 `FittedStar`, 1 `RefStar`) that the "correct" answers could be computed in advance with e.g. Mathematica.
 They are updated when changes to the algorithm warrants, e.g. a change to outlier rejection that affects the final chi2/ndof for some of the testdata.
-The `pa1` (see LPM-17 table 14) and relative/absolute rms (related to, but not the same as LPM-17's AM1 in table 18) values should be considered regression tests: algorithmic changes shouldn't result in larger values, but could produce smaller ones.
-They are "measured value must be less than" tests, and the calculations are in `lsst.jointcal.utils.JointcalStatistics.py`.
 
-More sophisticated testing of jointcal's validity is handled by `validate_drp` on much larger datasets that have multiple full-focal plane visits.
+More sophisticated testing of jointcal's validity is handled by `faro` on much larger datasets that have multiple full-focal plane visits.