For some specific issues like clipping sure there are objective methods, and encoders use objective quality levels (so you don't target bitrate), and if it's not a tricky "subjective human experience can't be objective" question - the method to test is do A/B testing and see that humans can't detect a difference. That's audio transparent encoding. And since you don't care about other people for your personal phone use case, you can do this test yourself to detect the encoding quality level that's transparent to you.

And your examle doesn't make sense. You compare reencoded to the original, not to some 3rd sample. The issue here is whether to store large flacs or their smaller reencoded lossy variants. If you get a better quality flac, then you'll need to do another encoding to get a better lossy version