Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Clean up kubelet eviction events

**What this PR does / why we need it**:
This makes eviction events better.  
* Exceeding container disk limits no longer says "node was low on X", since the node isn't actually low on a resource.  The container limit was just exceeded.  Same for pods and volumes.
* Eviction message now lists containers which were exceeding their requests.  This is an event from a container evicted while under memory pressure: 
`reason: 'Evicted' The node was low on resource: memory. Container high-priority-memory-hog was using 166088Ki, which exceeds its request of 10Mi.`
* Eviction messages now displays real resources, when they exist.  Rather than `The node was low on resource: nodefs`, it will now show `The node was low on resource: ephemeral-storage`.

This also cleans up eviction code in order to accomplish this.  We previously had a resource for each signal: e.g. `SignalNodeFsAvailable` mapped to the resource`nodefs`, and `nodefs` maps to reclaim functions, and ranking functions.  Now, signals map directly to reclaim and ranking functions, and signals map to real resources: e.g. `SignalNodeFsAvailable` maps to the resource `ephemeral-storage`, which is what we use in events.
This also cleans up duplicated code by reusing the `evictPod` function.  It also removes the unused signal `SignalAllocatableNodeFsAvailable`.

**Release note**:
```release-note
NONE
```

/sig node
/priority important-longterm

/assign @dchen1107 @jingxu97