Check role before including host in inactive host list

lock to avoid concurrent map writes

Maxunavailable reset should check number on inactive hosts

error out if service options are not found

it's unexpected to see empty service options, but node plan shouldn't be generated
in those scenarios

Do not wipe the actual Docker error to show in log

Attempt upgrade on NotReady hosts during cluster upgrade

Fix formatting

Return correct error if maxUnavailable hit for control

Accept label to ignore nodes during upgrade

Stop vendoring KDM and use data.json

RKE does a cluster scan to find the unreachable hosts, and if that number
is same as or exceeds maxUnavailable, upgrade won't proceed.
This commit introduces a label users can provide for their nodes so they
don't get counted as unavailable and are excluded from upgrade.
This commit also includes a couple of bug fixes

Problem: Docker error is shown as nil in log message
Root cause: err got redefined in loop and passed value was nil
Solution: do not redefine err in loop

Fix always show certificates are about to expire issue

Parse updateStrategy, replicas and autoscaler fields

Change RKE upgrade logic for zero downtime

Determine etcd s3 snapshots by s3 config presence

change the apiVersion of eventratelimit

Drop the log level to Debug since rancher can call this frequently

vendor update

Drop the log level to Debug since rancher can call this everytime worker connects back every ~2 mins