```
verifying github.com/ipfs/go-datastore@v0.0.1/go.mod: checksum mismatch
        downloaded: h1:bYmHO9fuKO1Ca7dpdDBWQl0mndy5b0HFqSJjGlNYtzs=
        go.sum:     h1:d4KVXhMt913cLBEI/PXAy6ko+W7e9AhyAKBGh803qeE=
```

```
verifying github.com/ipfs/go-datastore@v0.0.1: checksum mismatch
        downloaded: h1:u3lrOIBAhGS5prBa94V89LJNMWlEsO2+zgFlEGTl0E4=
        go.sum:     h1:AW/KZCScnBWlSb5JbnEnLKFWXL224LBEh/9KXXOrUms=
```

Go 1.11 has some weird go module checksum issues.

Don't start the control loop till we start the queue. The _in_ queue will still
work, we just won't start processing peers from it until we start the control loop.

fixes libp2p#311

go-libp2p-peerstore now has a TempAddrTTL of 2 minutes

This is just extra work as we write through anyways.

We only access it from a single goroutine.

1. Don't be n^2.
2. Don't bother walking the cache, just drop it.

batches deletes

Add initial set of metrics exposed via OpenCensus.

Otherwise, we'll delete everything.

Motivation: Walking the datastore can take time and currently blocks
adding/removing providers.

We need to do this in the same goroutine to avoid some logical races.

Long-running queries can build up large error sets that we never actually use.
This is exacerbated by libp2p/go-libp2p-swarm#115.

fixes libp2p/go-libp2p-swarm#119

* Expose metrics views in a slice

* Rename Views to DefaultViews

This communicates that these aren't the only views that are possible, or that they should be considered mandatory.

When the routing table is empty, `Run` would fail but _not_ close the
process (leaking some query goroutines). This patch fixes this in multiple
places by:

1. Not starting queries with no peers.
2. Failing queries with no peers earlier.

Allocate them as-needed and use a pool.

Work towards libp2p#322.

Deal with renamed tags.

Really deal with the deleted tag issue.

This was causing us to build up a bunch of background goroutines. Technically,
we should be resetting all these streams when the DHT stops. However, we don't
actually unregister stream handlers from the host so, really, nothing will
actually stop till the host stops. When it does, it'll kill all the streams
anyways.

This is a -u=patch update

Update msgio to latest version
Use max size in msgio readers
Fix error handling in reads