See https://24ways.org/2016/css-writing-modes/ and http://labs.jensimmons.com/#writing-modes
Hyatt says that we currently have the infrastructure to do this quite easily, so the bulk of the work would be to hook up the new properties.
Hyatt says that these would probably best be implemented with two independent properties: one for line direction which could be rl, tb, and bt, and one for line orientation, which could be normal or flipped. Then writing-mode would be a shorthand for combining these two.
Hyatt says 1-2 days of work
'sideways-rl' and 'sideways-lr' are at-risk in the latest CR spec https://www.w3.org/TR/css-writing-modes-3/#block-flow and have been removed from the editor's draft https://drafts.csswg.org/css-writing-modes-3/#block-flow which lead me to say that supporting 'sideways-rl' and 'sideways-lr' is NOT required in order to claim compliance with CSS3 Writing Modes spec. In all fairness, P2 in this bug report does not seem justified in my opinion.
Implementation of writing-mode: sideways-rl and sideways-lr is still needed. People are regularly trying to achieve the desired result (and even teaching others to do it) using the other writing-mode properties, even though such an approach only works in extremely limited circumstances and may lead to breakage unless carefully controlled. CSS Writing-modes 4 is in CR, and was this feature was split from level 3 in order to release what was already available for the benefit of CJK users, however the idea was that these values would be also implemented soon. It's a feature that plenty of people will benefit from, so can we move up the implementation if it's so straightforward? (It's already implemented by Gecko.)