grpcproxy: Disable fast fail on lease grant call to cluster

[yudai]

Hello,

Problem Observed
----------------

When there is no etcd process behind the proxy,
clients repeat resending lease grant requests without delay.
This behavior can cause abnormal resource consumption on CPU/RAM and
network.

Problem Detail
--------------

`LeaseGrant()` uses a bare protobuf client to forward requests.
However, it doesn't use `grpc.FailFast(false)`, which means the method returns
an `Unavailable` error immediately when no etcd process is available.
In clientv3, `Unavailable` errors are not considered the "Halt" error,
and library retries the request without delay.
Both clients and the proxy consume much CPU cycles to process retry requests.

Resolution
----------

Add `grpc.FailFast(false))` to `LeaseGrant()` of the `leaseProxy`.
This makes the proxy not to return immediately when no etcd process is
available. Clients will simply timeout requests instead.

I'm wondering if we should fix the client instead of modifying the proxy.
Currently the retry mechanism in clientv3 doesn't have any delay, so this issue can happen in the future again if somebody implements a proxy without this (undocumented) knowledge. (I actually made the same mistake in my private proxy)

[heyitsanthony]

The client should probably have some rate limiting on the retry path, but this seems OK to have as well.

[yudai]

The client should probably have some rate limiting on the retry path, but this seems OK to have as well.

I'll create a PR for this later when I have time. Because current retry mechanism makes writing third party proxies a little bit harder.

[xiang90]

@yudai thanks. merging.