> I'm not a fan of aiming for 100% coverage

"100% coverage" is just a measure of line coverage and is highly useless to reason about.

For testing to become meaningful, you should have test data covering way more than that for the relevant parts.