Can self-report instruments of shoulder function capture functional differences in older adults with and without a rotator cuff tear?