Primary key creation fails with duplicate key violation even though there are no duplicate values

Steps to produce

- Created a local table t1(a int, b int);
- Insert 2 rows (1, 1), (1,2);
- Create distributed table
- Create primary key on column a, fails with duplicate error (a=1);
- Delete violating rows (delete from t1 where a = 1), now table is empty
- Create primary key on column a, fails with duplicate error (a=1);

I believe this is due to us preserving local values of a table although it is distributed.