a NonVoter node should never try to bootstrap (hashicorp#492)

This is a follow-up to hashicorp#483

Snapshot restore progress (hashicorp#490)

When restoring a snapshot (on startup, installed from the leader, or during recovery) the logs are extremely terse. There are typically bookend messages indicating that a restore is going to happen, and that it is complete, but there's a big dead space in the middle.

For small snapshots this is probably fine, but for larger multi-GB snapshots this can stretch out and can be unnerving as an operator to know if it's stuck or still making progress.

This PR adjusts the logging to indicate a simple progress log message every 10s about overall completion in bytes-consumed.

Leadership transfer gaps (hashicorp#487)

* setLeadershipTransferInProgress before starting the transfer go routine to avoid a race.

* do not transition to follower state in a middle of a leadership transfer

a NonVoter node should never be able to transition to a Candidate sta…

…te (hashicorp#483)

* only voters can transition to a `Candidate` state

* clarify log message and remove not used func `inConfig`

* add tests to make sure leader transition happen as expected when the `HeartbeatTimeout` is reached

* fix race

add missing import

add retraction to version 1.1.3

Forbid removed node from being a `Candidate` (hashicorp#476)

* modify `TestRaft_RemoveFollower` to check removed node state

* add test to verify that a removed node is not able to vote

* do not transition to `Candidate` state if not part of stable configuration

* add some comments to the tests

* do not return if not transitioning state

* add wait loop to test

* remove test related to removed node voting

* check `inConfig` instead of `hasVote`

* updating warn log message to be more accurate

Merge pull request hashicorp#465 from hashicorp/dnephin/fix-panic-in-…

…decodePeers

Handle installSnapshot decodePeers error without a panic

Update Changelog (hashicorp#464)