When explicit -mavx etc are added, Clang generates vectorized ADDs as well:

(gdb) disas /m foo2
Dump of assembler code for function foo2:
23              return ((addr[0] + addr[1] + addr[2] + addr[3] + addr[4] + addr[5]) == 0);
   0x0000000000000054 <+4>:     vpmovzxbd (%rdi),%xmm0
   0x0000000000000059 <+9>:     movzbl 0x4(%rdi),%eax
   0x000000000000005d <+13>:    vpshufd $0x4e,%xmm0,%xmm1
   0x0000000000000062 <+18>:    vpaddd %xmm0,%xmm1,%xmm0
   0x0000000000000066 <+22>:    vpshufd $0xe5,%xmm0,%xmm1
   0x000000000000006b <+27>:    vpaddd %xmm1,%xmm0,%xmm0
   0x000000000000006f <+31>:    vmovd  %xmm0,%ecx
   0x0000000000000073 <+35>:    add    %eax,%ecx
   0x0000000000000075 <+37>:    movzbl 0x5(%rdi),%edx
   0x0000000000000079 <+41>:    add    %ecx,%edx