Questions about SAM2 function merge progress in tracking annotation

Recently, we have observed the impressive tracking capabilities of the Segment Anything (SAM) series models. I am very interested in leveraging these models to assist with video annotation in Label Studio. However, it appears that the official installation documentation is currently broken, as reported [here](https://github.com/HumanSignal/label-studio-ml-backend/issues/818).

A community member has shared a working solution in this pull request: [#826](https://github.com/HumanSignal/label-studio-ml-backend/pull/826). In addition, there were issues related to frame propagation, which seem to have been addressed in this commit: [f9cf1e9](https://github.com/HumanSignal/label-studio-ml-backend/pull/776/commits/f9cf1e9f2b594423d3420fbc8249a01931f8ae10). However, neither of these fixes has been reviewed or merged into the main branch.

Given this situation, I would like to ask whether support for SAM 2.1 or SAM 3 in video tracking annotation is still part of the development roadmap. Alternatively, is there currently no plan to maintain or extend SAM-based video tracking support in Label Studio?

Thank you for your time and clarification.