is `dc_stop_io()` needed before `dc_backup_provider_new()` or not?

[r10s]

i am a bit confused - imu the gist of #4138 was to free the UI from the burden to care for start/stop IO.

i just wanted to clean up the corresponding code on iOS - but without stopping IO, creating a backuo provider fails with the error:

ERROR deltachat_repl > Failed to set up second device setup: Database export failed: cannot export backup, IO is running

i double checked that in the repl - send-backup does not work when connected.

or ist it still needed to stopIO manually and core will only take care that calls to startIO are void during backup transfer? (that would also be a win)

related to #4007

[link2xt]

Likely because #4007 was written independently, it does not use the call context.scheduler.pause which was only introduced in #4138. Now need to fix the backup transfer to use this call too and get pause guard rather than just checking that the scheduler is stopped. The pause guard is used in imex(), but not in the new functions. I added pause guard to get_backup() here: #4177 Need to also find a place where it should be in the sender side.

[gerryfrancis]

Maybe related: #2813

[link2xt]

Maybe related: #2813

Yes, #2813 is very likely fixed by #4138.