The procedure to simulate recordings from a scout is documented here:
It is simpler than what you seem to have done here.
The leadfield is computed only once, then if you want to apply the signal to different dipoles, you create a different scout (with only one vertex) and repeat the procedure.
The process "Simulate recordings from scouts" creates both the source maps (with all the values set to zero except for the simulated dipole), and the corresponding recordings (multiplication of the sources with the forward model).