Memcached plugin fails all servers if one fails

[phemmer]

Bug report

Relevant telegraf.conf:

[[inputs.memcached]]
    servers = [ "localhost:11210", "localhost:11211" ]

System info:

telegraf 1.0.0 beta 2 commit c046232

Steps to reproduce:

telegraf -config teletest.conf -test

Expected behavior:

* Plugin: memcached, Collection 1
2016/07/03 14:27:24 dial tcp 127.0.0.1:11210: getsockopt: connection refused
> memcached,host=whistler,server=localhost:11211 bytes=0i,bytes_read=7i,bytes_written=0i,cas_hits=0i,cas_misses=0i,cmd_get=0i,cmd_set=0i,conn_yields=0i,connection_structures=11i,curr_connections=10i,curr_items=0i,decr_hits=0i,decr_misses=0i,delete_hits=0i,delete_misses=0i,evictions=0i,get_hits=0i,get_misses=0i,incr_hits=0i,incr_misses=0i,limit_maxbytes=67108864i,threads=4i,total_connections=11i,total_items=0i,uptime=21i 1467570428000000000

^ plugin should print error to STDERR, and then continue with the next server.

Actual behavior:

* Plugin: memcached, Collection 1
2016/07/03 14:27:24 dial tcp 127.0.0.1:11210: getsockopt: connection refused