To have some control over my data and for the fun of it. With money? Selfhosting to me is a hobby and it has also taught me a lot, some of which has been helpful on my career.
I don't have a straight up list to offer you, but https://lemmy.dbzer0.com/u/db0 has made a software for looking up instances with high amount of users but low amount of posts called lemmy-overseer. https://github.com/db0/lemmy-overseer
On github you can find link to their running instance with API documentation.
I have not personally used it, so this is as far as I can point you :)
Lemmy uses ActivityPub so even in a case where Lemmy would for whatever reason decide to use software You do not like, as long as other application that support the standard exists you'll be able to access the same content. As a side note, I'm surprised you linked directly to YT instead of an Invidious instance which would offer more privacy and is open source. Sure the data is still fetched from a service, but some is better than none, right?
For your first point, I post this from a server that uses non-free components for firmware and microcode. Also it might have non-libre software as it does not run a libre distribution. Whilst I do agree agree with the idea of opting to use libre software, sometimes in the current market it is not feasible. Technically one could colocate fully libre machines to use as servers, but when running something as a hobby it's likely to exceed the budget one is willing to spend.
I do agree with with preferring a selfhosted software solution to services wherever feasible, even better when libre. "Funnily" enough even some libre software end up including non-free software like in the case of microcode in Libreboot. Makes sense to run updated code which has security vulnerabilities missing compared to the unpatched code in the CPU that will be running regardless of ones preference.
Now on the linked article about open source missing the point. I can see where the article is coming from, but I guess some visibility is better than none? Sure I'm sure everyone would prefer to not make a compromise but what can You do in case no libre option exists?
I disagree with the last note. While that seems to be the case especially with free closed source services and current tracking practices nowadays, I do not believe that proprietary software is the cause, but rather its just one way to obscure what is happening in the background. For example when I buy a single player offline video game, I don't see myself as the product.
Seems like you are building Lemmy from source as it has "build:" specified. I personally use "image: dessalines/lemmy:0.17.4" to avoid building Lemmy from source.
True. In my case I was luckily able to root it and install YouTube Adfree from homebrew channel but apparently LG has patched most of the exploits in newest firmwares.
Secrets in password manager, documentation in NixOS configuration files.