SKS-2057: Bump CAPI to v1.6.3

[haijianyang]

Issue

Bump CAPI to v1.6.3

Change

Management Cluster: v1.25.x -> v1.29.x
Workload Cluster: v1.23.x -> v1.29.x

升级的组件：

• controller-runtime v0.16.5
• cert-manager v1.14.4
• controller-gen v0.13.0
• golangci-lint v1.55.2
• ginkgo v2.13.1
• Go 1.21.8

新增：

• IPAM: Promote IPAM types to v1beta1 (✨ promote IPAM types to v1beta1 kubernetes-sigs/cluster-api#9525)
• MachineSet: Adjust preflight check to allow kubelet version skew of 3 for clusters running v1.28 and above (✨ machineset: adjust preflight check to allow kubelet version skew of 3 for clusters running v1.28 and above kubernetes-sigs/cluster-api#9222)

Bug Fixes:

• CABPK: Certificate paths in cloud-init scripts should not use a platform-dependent path separator (🐛Certificate paths in cloud-init scripts should not use a platform-dependent path separator kubernetes-sigs/cluster-api#9167)
• clusterctl: Fix provider namespace secret not included in clusterctl move (🐛 fix provider namespace secret not included in clusterctl move kubernetes-sigs/cluster-api#9694)
• ClusterResourceSet: Requeue after 1 minute if ErrClusterLocked got hit ([release-1.6] 🐛 clusterresourceset: requeue after 1 minute if ErrClusterLocked got hit kubernetes-sigs/cluster-api#9788)
• KCP: Allow dropping patches KubeadmControlPlane KubeadmConfig (🐛 Allow dropping patches KubeadmControlPlane KubeadmConfig kubernetes-sigs/cluster-api#9698)
• KCP: Allow to drop useExperimentalRetryJoin field from KubeadmControlPlane.kubeadmConfigSpec (🐛 Allow unsetting useExperimentalRetryJoin field from KubeadmControlPlane kubernetes-sigs/cluster-api#9170)
• KCP: Fix KCP Controller reconcile always return error when workload cluster is unreachable (🐛 Fix KCP Controller reconcile always return error when workload cluster is unreachable kubernetes-sigs/cluster-api#9342)
• KCP: Requeue KCP object if ControlPlaneComponentsHealthyCondition is not yet true (🐛 requeue KCP object if ControlPlaneComponentsHealthyCondition is not yet true kubernetes-sigs/cluster-api#9032)
• Machine: Retry Node delete when CCT is locked (🐛 Retry Node delete when CCT is locked kubernetes-sigs/cluster-api#9570)
• MachineDeployment: Use regular random suffix for MachineSets, ensure max length 63 (🐛 MD controller: use regular random suffix for MachineSets, ensure max length 63 kubernetes-sigs/cluster-api#9298)
• MachineHealthCheck: Fix excessive trace logging in the machine health check controller (🐛 Fix excessive trace logging in the machine health check controller kubernetes-sigs/cluster-api#9419)

Test

参考 E2E