[Complexity] Optimize imult1616()

Bug description

The function imult1616() is currently using two operators for the 16x16 integer operation. However, the BASOP operator i_mult(v1, v2) could be used instead.

Word16 imult1616( Word16 x, Word16 y )
{
    assert( (int) x * (int) y < 32768 && (int) x * (int) y >= -32768 );
#ifdef PATCH
    return i_mult( x, y );
#else
    return extract_l( L_mult0( x, y ) );
#endif
}