This is my take on link chain sprockets. I think link chains as mechanical parts are a really intersting option for things like height adjustable desks etc. since the chains are really cheap and highly available in home improvement stores. They also can take high loads, however they are absolutely inappropriate for high speed or high precision jobs. They tend to have a decent amount of backlash too. Also they are quite noisy and non-linear (when you pull the chain with constant speed, the sprocket will rotate with non-constant speed and vice versa).
For horizontally arranged sprockets, you will need support rails made of a slippy material to prevent sagging (which will increase friction & noise).
To avoid need for a chain coupling, try spot-welding. I haven't experimented with polyethylene chains yet, will post results as available.
Before you start:
Especially for big gears ( = a lot of segments/teeth), the input parameters will require some try-and-error!
You should set cookiecutter_enabled = true and do a thin-walled test print.
Once your actual chain is running nicely (=without slipping AND without binding) on your test-printed sprockets, you set cookiecutter_enabled = false and enable shafting/hexnut/cavity as needed instead.
set do_preview = true
render preview and reconfigure parameters until you think the output might be ok.
the "interval" (=effective length of one link) parameter is most critical. To get it, stretch out about 1 Meter of your chain down on a nail in a wall and use a ruler to approximate it . i.E. from the top edge of the first link to the top edge of the 50th link (=49 links) you measure 933mm, then your interval value is 933/49 = 19,04
set do_preview = false, set do_middlesplit to 0,1 or 2 to get actual sprocket mesh. rendering may take a few minutes.