Hey there! Sorry you're having to deal with this. There's not necessarily one solution or any silver bullet, but there are many things you can try.
The biggest thing I can suggest is to raise your level of thinking from UX to the business metrics. By this I mean, rather than trying to optimize your product for "best UX", try to optimize it for "delivering most value to the customer." Creative people don't like to admit it, but sometimes the creatively optimal thing (good UX/good design/etc) doesn't always imply moving the needle of the business metrics. If you don't believe me, look at how much Amazon sells with their terrible UX. Ultimately the only thing that matters in a business is delivering value to the customer, and if the bad UX doesn't get in the way of that then you can sleep well knowing that your lead's bad UX decision didn't hurt anything.
I've found that thinking in these terms allows me to separate myself and let go of concerns like these. If the poor UX decisions of your manager are having a business impact then you can go find that data and put a dollar amount on it. For example, "I did some A/B tests and we got a 12% lower clickthrough rate with the button there." Your manager may not agree, but at least now you're talking about the customer, not about UX. If you can't produce that data then you still win because the business is unaffected by this mistake and you get to make your lead happy by doing it their way. Winning trust by doing bad UX is still winning trust.
Here's an example-by-story: I was once an engineer tasked with building an automated testing system. Automated tests are easy to understand and so everybody has opinions about how they should work. And so my system's design quickly got bogged down with folks giving the same sort of "i like it better that way" arguments that your lead is giving you. So, rather than engage with any of them, I picked a single battle in the design of my system, something that nobody could argue against, namely that tests should actually be test descriptions, so that the test system can work with declarative data and make decisions about the tests more easily. I showed how that would have a measurable impact on the business (having the test descriptions gives the test system flexibility so we can e.g. filter tests on the fly so that at desk testing times can be kept to a minimum, or easily move tests between BVT and Smoke, or whatever) which sidestepped all of those arguments.
If that doesn't float your boat, I like lostcolony's approach. One possible route of addressing it is to go through the hierarchy. As they mention it can be a tricky approach, you should avoid saying anything negative about your leads or explicitly asking anyone to change their behavior.
You could also try just having an earnest conversation with this person, expressing to them how their behavior makes you feel, and asking them what you should do to gain their trust and the ability to work autonomously. Try to make the conversation not about their behavior, but rather about what actions you can take. "What can I do to take more of these responsibilities on myself?" Then if you hold up your side of the deal, they're more likely to be accommodate you.
It may not be what want to hear, but getting someone else to change their behavior is one of the hardest things you may ever have to do in life and is often impossible. You might have to deal with the fact that this person will always do this. As lostcolony says, you can quit, but I see this as a last resort. You're just going to run into the same problems elsewhere, and it benefits you to learn to deal with the problems where you are.
Good luck!
Jorge
Edit: Looks like sizzle said exactly what I ended up saying but much shorter. And lostcolony's reply to sizzle brings up a good point: You have to present your case delicately, be careful not to make this person look bad, and if they don't respond to hard customer data, maybe it's time to find a team that does.