Any interval type with decidable equality is degenerate (everything ends up with a path to everything else). Having i0 and i10 be constant makes me suspect that the interval type likely has a decidable equality. This of course being on top of it missing a critical requirement to even call it an interval (transport).
I don't really understand why that section or the HIT talk is in one of the blog posts. All it seems to amount to is "yeah we can't do this." I suspect you will never be able to! There is a reason Cubical Type Theory adds the interval as a pretype with custom syntax, it is because the equality must become relevant and track exactly how paths apply. Without this relevant information I seriously doubt you will ever have any of those "possibilities."
I don't really understand why that section or the HIT talk is in one of the blog posts. All it seems to amount to is "yeah we can't do this." I suspect you will never be able to! There is a reason Cubical Type Theory adds the interval as a pretype with custom syntax, it is because the equality must become relevant and track exactly how paths apply. Without this relevant information I seriously doubt you will ever have any of those "possibilities."