So I've found something interesting ... If I replace THD_copy_dset_subs with the copy data set command PLUTO_copy_dset - it works. Data gets written to disk in real time, every TR, albeit the entire data set, and it gets bigger every TR. So I see that data being written to disk. So could there be some issues with THD_copy_dset_subs ?
Here are the output messages when trying to use THD_copy_dset_subs:
RT_sub_brick_id is [1, 0]
Trying to write data set rt._ch000_000000
RT_sub_brick_id is [1, 1]
** failure: THD_setup_mastery() returned 3
Trying to write data set rt._ch000_000001
** ERROR: EDIT_dset_items[90]: invalid input dataset
RT_sub_brick_id is [1, 2]
** failure: THD_setup_mastery() returned 3
Trying to write data set rt._ch000_000002
Thanks all.