Your technical details are about 10 years out of date.
The main point of a VPN isn't to circumvent security, it is to protect yourself from malicious actors. This could include those that want to track you for analytics you don't agree with. Major public VPNs are easy to block without DPI since they rely on DNS resolution. Preventing this blocking generally isn't a goal of a VPN service. Sure there are exceptions in VPNs designed to circumvent state censoring, but that isn't what Private Relay is for. DPI in general doesn't work well these days because certificate pinning and HSTS has become common. Technically you could trick HSTS in to thinking it has a legitimate certificate when doing DPI for machines that have an always-on VPN or never leave a corporate network, but that isn't common. It is mostly just used in highly regulated environments like within financial companies that are willing to live with the downsides. You generally can't use DPI without breaking much of the Internet. DPI used to be common on corporate networks before modern security practices, but most networks are primarily relying on technologies like OpenDNS these days. In practice, parental controls and content filtering only work on machines managed by IT (or a parent) that can install a plugin in the operating system. Nobody breaks long running connections to disrupt VPNs anymore. I recall seeing that 10-15 years ago before long polling became a common web technique. These days that would break many websites.