Make corrections suggested from code reviews.

Don't vecorized array operations that can't be aligned at runtime
such as:

    int badstride = 11;
    for (int i = 0; i < len; i+= badstride) {
      ia[i+0] = ia2[i+0];
      ia[i+1] = ia2[i+1];
      ia[i+2] = ia2[i+2];
      ia[i+3] = ia2[i+3];
      ia[i+4] = ia2[i+4];
      ia[i+5] = ia2[i+5];
      ia[i+6] = ia2[i+6];
      ia[i+7] = ia2[i+7];
    }