That is correct, when you modify a validation or batch calculation rule, all rows in that table will reset its validationStatus (a field we use to know what to pick up during evaluation) to "dirty" which means mark them as require validation or calculation depending on what rules you have.
The next evaluation will pick all the rows and either generate more errors or remove the existing errors if they are not longer valid.
For example:
You have a table with 10 rows and validation rule that return false. (which means all rows are errors) you run evaluate and now you get 10 error features. and all 10 rows says clean (no require validation). If you evaluate again it no-ops and no rows are evaluated. (performance thing)
If you change the validation to return true, we touch all 10 rows to make them require validation. so each row will be evaluated and the row errors will be deleted because the rule evaluate to true. you will end up with no errors and 10 rows that require no evaluation.
