Imagine you are uploading or downloading many files in parallel, but discover that one of your larger files is either a) wrong or b) not meant to be downloaded. you have a choice. kill your entire rclone session (possibily losing all in flight data if resume isn't possible) or simply be willing to waste the data on the "bad" file.
if one the other hand you could do
rclone rc list-transfers
- which would return some form of unique id that would be relatively uniqiue to identify existing transfers
and
rclone rc kill-transfer <id>
- which would kill the transfer specified
one could avoid this situation.
this would possibly be part of generic queue management discussed in other parts, i.e. how to add new files to the transfer queue, so as not just having to start a new parallel rclone run.