missed GROUP BY optimization on JOIN with simple equality condition

[smavros]

the GROUP BY common sub-expression elimination (CSE) optimizer is not acting upon

select f.n, b.n from f join b on f.n = b.n group by f.n, b.n;

the PLAN indicates two grouping columns even though they are going to have same values

project (                                    
| group by (                                 
| | join (                                   
| | | table("sys"."f") [ "f"."n" ],          
| | | table("sys"."b") [ "b"."n" ]           
| | ) [ ("f"."n") = ("b"."n") ]              
| ) [ "b"."n", "f"."n" ] [ "f"."n", "b"."n" ]
) [ "f"."n", "b"."n" ]                       

this is going to be probably tricky since the optimizer need to parse the subrelation operator's expression tree and understand the semantics

tested on default at b0f51c763c66 (mercurial changeset)