The 90 days disclosure you’re referencing, which I believe is primarily popularized by Google’s Project Zero process, is the time from when someone discovers and reports a vulnerability to the time it will be published by the reporter if there is no disclosure by the vendor by then.

The disclosure by the vendor to their users (people running Lemmy instances in this case) is a completely separate topic, and, depending on the context, tends to happen quite differently from vendor to vendor.

As an example, GitLab publishes security advisories the day the fixed version is released, e.g. https://about.gitlab.com/releases/2024/01/11/critical-security-release-gitlab-16-7-2-released/.
Some vendors will choose to release a new version, wait a few weeks or so, then publish a security advisory about issues addressed in the previous release. One company I’ve frequently seen this with is Atlassian. This is also what happened with Lemmy in this case.

As Lemmy is an open source project, anyone could go and review all commits for potential security impact and to determine whether something may be exploitable. This would similarly apply to any other open source project, regardless of whether the commit is pushed some time between releases or just before a release. If someone is determined enough and spends time on this they’ll be able to find vulnerabilities in various projects before an advisory is published.

The “responsible” alternative for this would have been to publish an advisory at the time it was previously privately disclosed to admins of larger instances, which was right around the christmas holidays, when many people would already be preoccupied with other things in their life.