--stop-timeout is not respected when changing number of running Users in distributed mode

[heyman]

Describe the bug

When running Locust distributed, and changing the number of simulated Users (e.g. if a new slave node connects), simulated users will be killed immediately without respecting --stop-timeout

Expected behavior

Running locust user should get a chance to finish their tasks within the specified stop-timeout.

Actual behavior

Running locust users are killed immediately.

Fix

I'm currently working on a fix for this.