FYI I've also updated #5613 few times to include your original commit with a could of improvements etc and have generally been keeping it rebased any time I update my project working branch.

That being said I'm still not sure if it's ideal for inclusion in mainline micropython as-is because the MR and/or underlying library does make a few assumptions that aren't ideal in all cases.

In my understanding, there are a few compromises in terms of ram use vs compression level that would be better customised on a per-project basis depending on the chip / resources available.

Perhaps this would all be much better as a dynamic C module where there could be a couple of versions available for different applications? I still haven't got my head around building/distributing dynamic C modulus properly though to know if this is a good/viable option.

Or maybe it is still better to not let perfection get in the way of good enough... I'm not too sure how much flash/ram is taken up by this current implementation, as a board #define to enable/disable as needed maybe this is good enough.

Superseded by #11905.