System backup continues when device switches to metered network

[grote]

While the backup is running, the system doesn't stop it on its own. It requests a backoff from time to time, we give it a high backoff, but it still continues as if nothing happened:

10-25 17:39:25.798 27278 27289 I BackupCoordinator: Request full backup time. Returned 3600000
10-25 17:39:25.800  1361 29957 I PFTBT   : Transport suggested backoff=3600000
10-25 17:39:25.801  1361 29957 V BackupManagerConstants: getBackupFinishedNotificationReceivers(...) returns 
10-25 17:39:25.801 27278 27289 I NotificationBackupObserver: Completed. Target: net.etuldan.sparss.floss, status: 0
10-25 17:39:25.801  1361 29957 I PFTBT   : Unbinding agent in net.etuldan.sparss.floss
10-25 17:39:25.802  1361 29957 W ActivityManager: Unbinding backup agent with no active backup
10-25 17:39:25.802  1361 29957 I PFTBT   : Initiating full-data transport backup of nodomain.freeyourgadget.gadgetbridge token: 725718313
10-25 17:39:25.811 27278 27289 I FullBackup: Perform full backup for nodomain.freeyourgadget.gadgetbridge.
10-25 17:39:25.815 27278 27289 I BackupCoordinator: Get backup quota for nodomain.freeyourgadget.gadgetbridge. Is full backup: true.

[grote]

A low hanging fruit would be to ourselves not request backup for those remaining 100-app chunks. Then whatever apps are missing would still get backed up and use bandwidth.

Another approach could be to return TRANSPORT_PACKAGE_REJECTED for all apps while we are on a metered network. Need to see how the system reacts to that. May need faking a successful @pm@ backup, if this trips up the system.