Maybe I'm misunderstanding you, but it sounds like your problems with your VPS and VPN come from having someone else hosting those services for you and restricting how you can use them instead of just doing them yourself. Things are indeed much more permanent (at least as much as you want them to be) when you actually do them yourself instead of rent them from someone.
> Would you mind sharing a tutorial if you’ve saved one somewhere
I would if I had one, but I mostly just googled the official installation instructions for any random component I wanted. No overall tutorials. Service-wise, I'm using Jellyseerr to discover content and take requests from friends/family, Radarr for sourcing movies, Sonarr for sourcing TV shows, Lidarr for sourcing music, Prowlarr for centralizing the configuration of the other *arrs, Sabnzbd as a usenet download client, rtorrent as a bittorrent client, and Jellyfin for consuming my library. You probably don't need all of those depending on what you're after, but you can just look up the instructions for only the components you want. And if you want to get content from usenet or private torrent trackers, you'll need the relevant accounts.
> what disk, CPU, and memory I am looking at for this?
Whatever you want honestly. You can run most of this on a toaster so long as you don't have unrealistic performance expectations for it. Obviously if you're planning to download 10TB worth of content, you're going to need at least that much disk space, and if you're planning to download it faster than HDDs can spin, you might need some SSDs. But most of these things are just downloaders and file managers and don't really need much more than network bandwidth to source the content and disk IO to put the content somewhere. Maybe 8GB of memory for some of the more bloated services (my whole server idles at ~5GB used, and it runs a lot of other junk) and to do really large indexer merges when searching for content, and however much hardware you need to meet your transcoding requirements if you're playing through Jellyfin or similar. You can stream source material without transcoding and consume basically no CPU from it, or you can enable transcoding with a decent enough CPU or pretty much any supported GPU. All up to you and your needs.
I'd recommend starting with pretty minimal hardware and adding more as you need it. My storage has quadrupled since I started doing this and at some point I tossed in an old GPU for transcoding.