Hi Alex
First, I think I would make up an L shape test piece to duplicate part surface, fit on component or simulate same conditions and spray. Do metallography on whole L shape cross-section to establish what's happening to the coating. If coating looks good then you may need to concentrate on improving the machining/grinding procedure.
Chances are that you are suffering from:
ID spraying problems with ventilation of dust and debris away from coating area, but much worse than that is the L shape surface.
Excuse my scribble, but I think you will get the picture

When spraying one surface, spray material bouncing off will form a very poor quality secondary deposit on the adjacent perpendicular face.
The only way I can suggest for curing this problem is to mask either the vertical or horizontal face and spray the other at the best angle you can get. Remove masking and spray again with optimal spray angle for uncoated area.