linkerd-cni renames kuberouter cni config without warning

Nobody has claimed this yet.

## Bug Report

### What is the issue?

We're using kube-router as a complete replacement for kube-proxy (https://github.com/cloudnativelabs/kube-router). Upon deployment, the linkerd-cni daemonset renames the 10-kube-router.conf file to 10-kube-router.conflist without warning. Since kube-router only looks for 10-kube-router.conf by default, this breaks the default install of kube-router. The problem isn't noticed until next time a kube-router pod restarts and it goes into a crash loop.

### How can it be reproduced?

Deploy linkerd-cni

### Logs, error output, etc
Relevant linkerd-cni pod log output:
```
Renaming /host/etc/cni/net.d/10-kuberouter.conf extension to .conflist
Removing CNI_OLD_CONF_PATH: /host/etc/cni/net.d/10-kuberouter.conf
Created CNI config /host/etc/cni/net.d/10-kuberouter.conflist
Done configuring CNI. Sleep=true
```
### Environment

- Kubernetes Version: v1.13.3
- Cluster Environment: kubeadm bare metal
- Host OS: CentOS7
- Linkerd version: stable-2.2.1

### Possible solution

In the linkerd-cni install instructions, document upfront needed changes to kube-router config (or warn in general about the change and possible ramifications on existing CNI setups that have an expectation of cni config file name).

Start with the linkerd-cni daemonset behavior that handles files under /host/etc/cni/net.d/ and review the linkerd-cni install instructions. Reproduce the deployment with kube-router and verify whether the .conf file is renamed to .conflist; done means the behavior is clearly warned about or the kube-router configuration remains functional.

Written by the indexing model from the issue text.