enhance the HA of inluxdb

[willzgli]

What would you like to be added:

Import influxdb-cluster in tkestack

At presenet, an opensource influxdb cluster solution comes out (refer to: https://github.com/chengshiwen/influxdb-cluster). It's to be an alternative to InfluxDB Enterprise.

influxdb cluster must be used with persistence storage, and tkestack already support nfs with nfs-subdir-external-provisioner
(https://github.com/tkestack/tke/tree/master/cmd/tke-installer/app/installer/manifests/charts/nfs-subdir-external-provisioner).
They will work together.

Integration plan:

1. upload influxdb-cluster images to registry.tke.com/library

2. downlaod infuxdb-cluster charts from https://github.com/influxtsdb/helm-charts/tree/master/charts/influxdb-cluster

3. modify contents in chart
   3.1 replace the image registry with registry.tke.com/library
   3.2 replace storageclass name with nfs-client-provisioner (this is the defalut name)
   3.3 change the replicas of data node to 3 (default is 2)
   3.4 make sure the svc name of data node as same as current influxdb svc

4. move the modified chart in https://github.com/tkestack/tke/tree/master/cmd/tke-installer/app/installer/manifests/charts/

5. add two options in https://github.com/tkestack/tke/blob/master/cmd/tke-installer/app/installer/types/types.go#L295, such as single-influxdb-mode and influxdb-cluster-mode, users can select installation mode.

6. install one of influxdb chart in installInfluxDBChart() ( https://github.com/tkestack/tke/blob/master/cmd/tke-installer/app/installer/installer.go#L1889)

7. If use influxdb-cluster, we should change the metrics replicas to 3 , this part of work is in platform-controller

Why is this needed:

influxdb running in single instance mode has the risk of service break down.