Confusion about removed task groups

### Nomad version
$ nomad version
Nomad v0.11.2+ent (7e7a895f627eaf94d1f2e63634b035df02a04be2

### Issue
When task groups are removed during the lifetime of a job, the removed task groups are still displayed in the CLI and are included in the job summary metrics but are not visible in the UI.

This gets messy quickly - the CLI/UI discrepancy is confusing but the real pain are the job summary metrics as they make alerting rather ... difficult.

Ideally, the UI and CLI should **always** give the same information and job summary metrics should **NOT** be generated for task groups that have been explicitly removed from a jobfile.

### Reproduction steps
1. Deploy the jobfile below containing two task groups.

```
$ nomad plan foo.nomad
+ Job: "foo"
+ Task Group: "first_group" (1 create)
+ Task: "redis" (forces create)

+ Task Group: "second_group" (1 create)
+ Task: "redis" (forces create)

Scheduler dry-run:
- All tasks successfully allocated.

Job Modify Index: 0
To submit the job with version verification run:

nomad job run -check-index 0 foo.nomad

When running the job with the check-index flag, the job will only be run if the
job modify index given matches the server-side version. If the index has
changed, another user has modified the job and the plan's results are
potentially invalid.
```

After deploying the job :

- `nomad job status foo` will display both groups
- the UI will display both task groups
- job summary metrics will be generated for both task groups.

2. Once deployed, remove the task group `second_group` from the jobfile and run the job again.

```
$ nomad plan foo.nomad
+/- Job: "foo"
Task Group: "first_group" (1 in-place update)
Task: "redis"

- Task Group: "second_group" (1 destroy)
- Task: "redis" (forces destroy)

Scheduler dry-run:
- All tasks successfully allocated.

Job Modify Index: 798339
To submit the job with version verification run:

nomad job run -check-index 798339 foo.nomad

When running the job with the check-index flag, the job will only be run if the
job modify index given matches the server-side version. If the index has
changed, another user has modified the job and the plan's results are
potentially invalid.
```

- `nomad job status foo` will still display both groups
- the UI will only display the remaining, active task group
- job summary metrics will still be generated for both task groups.

### Job file (if appropriate)
```
job "foo" {
datacenters = ["dc1"]

group "first_group" {
task "redis" {
driver = "docker"

config {
image = "redis:3.2"

port_map {
db = 6379
}
}

resources {
cpu = 500
memory = 256

network {
mbits = 10
port "db" {}
}
}
}
}

group "second_group" {
task "redis" {
driver = "docker"

config {
image = "redis:3.2"

port_map {
db = 6379
}
}

resources {
cpu = 500
memory = 256

network {
mbits = 10
port "db" {}
}
}
}
}
}
```

Reproduce the lifecycle described with the provided Nomad job file, comparing `nomad job status foo`, the UI, and job summary metrics before and after removing `second_group`. Trace the paths responsible for task-group status and summary metrics; done means removed groups no longer appear in CLI output or metrics and the CLI and UI agree.

Written by the indexing model from the issue text.