It removes the has_part rule for treats. Review of both results and the cases that led to the role pointed towards this rule not being helpful.
Turned on within-result set promiscuity filtering.
To test, I ran 4 "treats" queries in dev and on a local version running this PR, and compares the ranks of expected results from the sprint 4 test suite, all with 60 minute query and KP timeouts.
Its difficult to totally nail down what's going on because results vary due to timeouts, but I think that this looks at worst neutral.
This PR makes two changes:
To test, I ran 4 "treats" queries in dev and on a local version running this PR, and compares the ranks of expected results from the sprint 4 test suite, all with 60 minute query and KP timeouts.
Its difficult to totally nail down what's going on because results vary due to timeouts, but I think that this looks at worst neutral.