Holding that where employer's interchange data was represented in aggregate form rather than as a percentage of total employees, "we have no basis for doubting the Board's conclusion that the amount of employee interchange . . . is either `not substantial' or is `minimal'"