Update ClusterRoleBinding to rbac.authorization.k8s.io/v1

[master] Prevent panic seen intermittently while logging in with SAML

refactor k3s to support swappable components

Update rke

* [2.5] switch to upstream saml library (#30858)

* switch to upstream saml library

* fixing go sum

* go mod tidy

[Forwardport] Restricted admin role changes

[master] adding back v1 node delete

Add support to docker 20.10 and set it as default install script

Updated telemetry version

Change dev version to be 2.6

We need not pull the memberof attribute of a group when listing groups during user login

[Forwardport] Enable agent image to be pulled from private registry

Remove kubeversion check

[master]Allow k8s 1.20 in rancher helm chart

[master] temporarily move charts branch to alt

[master] removing cluster agent check before cluster is ready

Add rke2 airgap deployment methods

[Validation] - Add trace logging to single node and HA installs for automation

Skip ServiceAccount tokens on secrets sync

[master] Update machine to v0.15.0-rancher50

Added etcd snapshot timeout parameter

The most significant part are the changes to syncing. The syncing
algorithm more carefully crawls the cluster spec to only sync the parts
that are not nil in the config. This is useful because it allows
tracking of the Rancher managed launch template in the upstream spec,
but this will not sync this template to the spec (the launch
template will be nil because it is not provided by the user).

The rancher managed launch template ID and template versions for
managed node groups are also tracked in status for updates.