Steps to Reproduce (get latest from master):

1. add a process.crash() somewhere into workbench code that executes after a timeout
2. run scripts/code.sh --crash-reporter-directory <path to empty folder>
3. wait for the crash
4. check the output of the crash folder

=> no dmp file exists (but other files that seem to indicate the crash directory is picked up)

This works fine when triggering a crash on the main side. So I wonder if we are doing something wrong with how we setup the crash reporter.