I haven't documented it as it's only in the develop/standalone branches, but you just set the parameter ('runtime', 'resume') = 'True' and it will do the rest, as long as the output file name is the same. So should be able to just use the same analysis id and it will work.
We will also need to add a --resume parameter to the run-cosmosis function which takes in the path to the output_* dir to resume from.
Notes from @joezuntz
We will also need to add a
--resume
parameter to the run-cosmosis function which takes in the path to theoutput_*
dir to resume from.