GH-30135 added an expensive ABC subclass check for an uncommon case (MappingView) before fast subclass checks for 2 common cases (list and tuple), resulting in a ~10% performance regression for our pprint benchmarks. There's also some evidence that this may have also made our PGO profile worse on some builds, since test_pprint is part of the profiling task (so we're spending a good chunk of that time hitting the newly-added ABC subclass-checking machinery).

This just switches the order of the branches so the cheap, common case is checked before the expensive, uncommon one.

• Issue: Teach pprint about dict views #90117