* mempool: Store txs by hash inside of cache

This allows for large cachesizes, without fear of the memory
consumption growing rapidly.

* (squash this) rename hashedTx -> txHash

* mempool: Keep cache hashmap and linked list in sync

This removes bugs with the linked list being full, but hashmap empty

* address PR comments

* switch clist back to list

* libs: Remove usage of custom Fmt, in favor of fmt.Sprintf

Closes #2193

* Fix bug that was masked by custom Fmt!

This PR changes ABCI time format from int64 (Unix seconds) to WKT (WellKnownType) google.protobuf.Timestamp.

Refs #1857

Reasons:

better precision
standard DT for proto

* update Gopkg.lock
* [makefile] remove extra grep
    - go list excludes vendor by default now
* proto3 timestamp
* [docs/abci-spec] note about serialisation format
* make time non-nullable

* #1920 try to fix race condition on proposal height for published txs

- related to create_empty_blocks=false
- published height for accepted tx can be wrong (too low)
- use the actual mempool height + 1 for the proposal
- expose Height() on mempool

* #1920 add initial test for mempool.Height()

- not sure how to test the lock
- can the mutex reference be of type Locker?
-- this way, we can use a "mock" of the mutex to test triggering

* #1920 use the ConsensusState height in favor of mempool

- gets rid of indirections
- doesn't need any "+1" magic

* #1920 cosmetic

- if we use cs.Height, it's enough to evaluate right before propose

* #1920 cleanup TODO and non-needed code

* #1920 add changelog entry

This reverts commit d02c5d1e30.

Refs #646

Fixes #1762#issuecomment-398940107

This reverts commit 96a3502

- move prometheus metrics into internal packages
- *Option structs
- misc. format changes

to better handle zero cache size

* updateToState exits early if the state isn't new, which happens after
* fast syncing. This results in not sending a NewRoundStep message. The mempool
* reactor depends on PeerState, which is updated by NewRoundStep
* messages. If the peer never sends a NewRoundStep, the mempool reactor
* will think they're behind, and never forward transactions. Note this
* only happens when `create_empty_blocks = false`, because otherwise
* peers will move through the consensus state and send a NewRoundStep
* for a new step soon anyways. Simple fix is just to send the
* NewRoundStep message during updateToState even if exit early

no longer relevant, I guess. since ABCI only defines 0 (success) code.

Refs #345

Refs #345

https://github.com/tendermint/tendermint/issues/1539#issuecomment-387240606

BEFORE:

```
./tm-bench -c 5 -r 1000 127.0.0.1:46657
Stats          Avg       StdDev     Max
Txs/sec        1826      843        2744
Blocks/sec     1.100     0.300      2
```

AFTER:

```
./tm-bench -T 30 -c 5 -r 1000 127.0.0.1:46657
Stats          Avg       StdDev     Max
Txs/sec        6120      1970       9776
Blocks/sec     1.000     0.000      1
```

* remove accidental binary

* docs: s/Dummy&dummy/KVStore&kvstore/g

* glide update to abci

* update abci import paths

* dummy begone, hello kvstore

* RequestInitChain needs genesisBytes

* glide update

1. remove pointer
2. add Quit() method to Service interface

https://github.com/tendermint/tendermint/pull/1173#issuecomment-363173047

https://play.golang.org/p/gN21yO9IRs3

```
func waitWithCancel(f func() *clist.CElement, ctx context.Context) *clist.CElement {
	el := make(chan *clist.CElement, 1)
	select {
	case el <- f():
```
will just run f() blockingly, so this doesn't change much in terms of behavior.

Leaking goroutine:
```
114 @ 0x42f2bc 0x42f3ae 0x440794 0x4403b9 0x468002 0x9fe32d 0x9ff78f 0xa025ed 0x45e571
```

Explanation:
it blocks on an empty clist forever. so unless theres txs coming in,
this go routine will just sit there, holding onto the peer too.
if we're constantly reconnecting to some peer, old instances are not
garbage collected, leading to memory leak.

Fixes https://github.com/cosmos/gaia/issues/108
Previous attempt https://github.com/tendermint/tendermint/pull/1156

NOTE we only updating for tmlibs and abci

if we call it after, we might receive a "fresh" transaction from
  `broadcast_tx_sync` before old transactions (which were not
  committed).

Refs #1091

```
Commit is called with a lock on the mempool, meaning no calls to CheckTx
can start. However, since CheckTx is called async in the mempool
connection, some CheckTx might have already "sailed", when the lock is
released in the mempool and Commit proceeds.

Then, that spurious CheckTx has not yet "begun" in the ABCI app (stuck
in transport?). Instead, ABCI app manages to start to process the
Commit. Next, the spurious, "sailed" CheckTx happens in the wrong place.
```

Include overview and configuration options.