Please check which Operator ID you use for Runtime registration. It seems there is a wrong one in place or it is set up incorrectly. If you need to try with another one - delete RPAStartup.info file under %appdata%/OpenSpan and try to start Runtime once again.
Operator Id's access group should have PEGARULES:RoboticAdministrator role.
Please also check if you have set up pyGetAccessGroupForRobotByWorkGroupdecision decision table.
The registration is OK and I can see the name of M3 machine as an operatorID I know these points about deleting RPAStartup.info to enter credentials another time and about the rule pyGetAccessGroupForRobotByWorkGroupdecision I can asume it is correct because on the contrary the registration of the robot would be wrong.
I have done the following test: I have load manually the openspan solution in the runtime and on creating a work case the assignation is queued and executed correctly by the M3 runtime.
I am thinking it is possible this happens because the robot manager version is the 5. I have seen in robot manager there is a dropdownlist to change the status of the queue of workgroup with options "running" and "stop", could it be reason? may I change the status of this queue to "running"?
I believe workgroup queue options affect only ability to send work items to queues, but it is not related to package assignments.
I'm able to reproduce the issue if I delete "AutomationPackageManagement:RuntimeUser" role from the access group. Could you please double check that this role is in place? It allows to retrieve a package.
If it still doesn't work, I don't have other ideas - maybe someone else can help.