Eating an animal at least ostensibly has positive value for the people doing so. However, there are plenty of forms of "animal testing" that confer zero positive value. For instance, testing the wrong compound or inserting the wrong implant confers zero benefit. Having improper controls, "testing" nonsensical theories, repeating stale results poorly, inadequate data collection, etc. are just a few ways a test procedure can be totally useless or even actively harmful.
This also ignores one of the other aspects of animal testing which is as a dry run or rehearsal for actual application. You do it right in animals so you are practiced at doing it right for when you need to do it right in humans. "Oh yeah, we royally screwed up in every rehearsal, but we will nail it in production." is not an acceptable approach. You look at the care taken during their practiced procedures on less critical subjects to determine if their practiced procedure is adequate for more critical subjects. A process that kills far more test subjects than others or achieves middling results relative to resource expenditure or that treats subjects as disposable for "advancing science" is not a process fit for human subjects. Assuming ingrained cultural process deficiencies will magically disappear when using changing subjects is foolish.
These are just some of the reasons why people eating a ridiculous number of animals does not and should not waive our invalidate concerns about animal testing procedure.