Update flannel from 0.6.2 to 0.8.0

Removing run_once in these tasks so that etcd ca certs get propogated…

Add etcd metrics flag

add newline after expanding user information

Fix enforce-node-allocatable option

Closes #1228
pods is default enforcement

see https://kubernetes.io/docs/tasks/administer-cluster/reserve-compute-resources/
add

update

without this etcd ca certs dont exist on worker nodes causing calico to fail

[terraform/openstack] fixed mistake in README.md

* Enable weave seed mode for kubespray

* fix task Weave seed | Set peers if existing peers

* fix mac address variabilisation

* fix default values

* fix include seed condition

* change weave var to default values

* fix Set peers if existing peers

set loadbalancer_apiserver_localhost default true

"failed_when: false" and "|succeeded" checks for registered vars

Template out known_users.csv, optionally add groups

Custom `etcd_data_dir` saves etcd data to host, not container

to match this https://github.com/kubernetes-incubator/kubespray/blob/master/roles/kubernetes/node/tasks/main.yml#L20
and the documented behaviour in HA docs 
related to #1456
@rsmitty

higher the timeouts for draining nodes while upgrading kubernetes version

Adding recursive=true for rkt kubelet dir

Kubernetes upgrade to 1.6.7

Add more instructions to setting up AWS provider

fix some typos in HA doc

fix typo 'on' > 'one'

Give more control over IPIP, but with same default behaviour

add configurable parameter for etcd_auto_compaction_retention

Fixes #1434

add vsphere cloud provider doc

fix typo

make sure every instance is a node if user changed defaults

Remove deprecated 'enable-cri' flag in kubernetes 1.7

fix azure kubernetes port to 6443

[terraform/openstack] README.md Guide expanded