Skipping Update Stanza when task group is renamed

Nomad seems to only honor the job's `update` stanza when task group details change, but not when the task group layout structure changes. For example, if a task group is renamed, Nomad will immediately shutdown all current allocations and start new ones and potentially cause a minor outage.

The reason is that generic scheduler (and reconciler) considers the updates for each task group independently. If a job renames a task group, the scheduler will treat the change as a task group removal and insertion.

We should consider documenting this explicitly if it's not documented already. Also, we can consider having max_parallel apply globally to all task groups, so we would only shutdown one allocation at a time

### Reproduction steps

Run the following commands
```sh
# run the first job, command reports 3 allocations
nomad job run ./canary-test.nomad
# update env-var value - see a single canary job starting
cat ./canary-test.nomad | sed 's|"val1"|"val2"|g' | nomad job run -
# check that there is a single running deployment with a single canary allocation
nomad job status example
# now submit the job but with group name change, note
cat ./canary-test.nomad | sed 's|"cache-0"|"cache-1"|g' | nomad job run -
nomad job status example
```

My sample output

```sh
$ nomad job run ./canary-test.nomad
==> Monitoring evaluation "0cf9bedf"
Evaluation triggered by job "example"
Allocation "67f44ee8" created: node "624b8edb", group "cache-0"
Allocation "d3764a91" created: node "624b8edb", group "cache-0"
Allocation "60110964" created: node "624b8edb", group "cache-0"
Evaluation within deployment: "03b744f3"
Allocation "d3764a91" status changed: "pending" -> "running" (Tasks are running)
Allocation "60110964" status changed: "pending" -> "running" (Tasks are running)
Allocation "67f44ee8" status changed: "pending" -> "running" (Tasks are running)
Evaluation status changed: "pending" -> "complete"
==> Evaluation "0cf9bedf" finished with status "complete"

$ # update env-var value - see canary
$ cat ./canary-test.nomad | sed 's|"val1"|"val2"|g' | nomad job run -
==> Monitoring evaluation "26a19a5e"
Evaluation triggered by job "example"
Evaluation within deployment: "2a5cf58c"
Allocation "0f98a673" created: node "624b8edb", group "cache-0"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "26a19a5e" finished with status "complete"

$ # notice deployment has one canary
$ nomad job status example
ID = example
Name = example
Submit Date = 2020-08-09T20:14:57-04:00
Type = service
Priority = 50
Datacenters = dc1
Namespace = default
Status = running
Periodic = false
Parameterized = false

Summary
Task Group Queued Starting Running Failed Complete Lost
cache-0 0 0 4 0 0 0

Latest Deployment
ID = 2a5cf58c
Status = running
Description = Deployment is running but requires manual promotion

Deployed
Task Group Promoted Desired Canaries Placed Healthy Unhealthy Progress Deadline
cache-0 false 3 1 1 0 0 2020-08-09T20:24:57-04:00

Allocations
ID Node ID Task Group Version Desired Status Created Modified
0f98a673 624b8edb cache-0 1 run running 8s ago 7s ago
60110964 624b8edb cache-0 0 run running 1m8s ago 57s ago
67f44ee8 624b8edb cache-0 0 run running 1m8s ago 57s ago
d3764a91 624b8edb cache-0 0 run running 1m8s ago 57s ago

$ # now submit the job but with group name change
$ cat ./canary-test.nomad | sed 's|"cache-0"|"cache-1"|g' | nomad job run -
==> Monitoring evaluation "24838f8b"
Evaluation triggered by job "example"
Allocation "f2ad41f7" created: node "624b8edb", group "cache-1"
Allocation "7165ad7e" created: node "624b8edb", group "cache-1"
Allocation "9c7e5681" created: node "624b8edb", group "cache-1"
Evaluation within deployment: "9b4d6fef"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "24838f8b" finished with status "complete"
$ nomad job status example
ID = example
Name = example
Submit Date = 2020-08-09T20:15:49-04:00
Type = service
Priority = 50
Datacenters = dc1
Namespace = default
Status = running
Periodic = false
Parameterized = false

Summary
Task Group Queued Starting Running Failed Complete Lost
cache-0 0 0 0 0 4 0
cache-1 0 0 3 0 0 0

Latest Deployment
ID = 9b4d6fef
Status = running
Description = Deployment is running

Deployed
Task Group Desired Placed Healthy Unhealthy Progress Deadline
cache-1 3 3 0 0 2020-08-09T20:25:49-04:00

Allocations
ID Node ID Task Group Version Desired Status Created Modified
f2ad41f7 624b8edb cache-1 2 run running 7s ago 6s ago
7165ad7e 624b8edb cache-1 2 run running 7s ago 6s ago
9c7e5681 624b8edb cache-1 2 run running 7s ago 6s ago
0f98a673 624b8edb cache-0 1 stop complete 59s ago 6s ago
60110964 624b8edb cache-0 0 stop complete 1m59s ago 6s ago
67f44ee8 624b8edb cache-0 0 stop complete 1m59s ago 6s ago
d3764a91 624b8edb cache-0 0 stop complete 1m59s ago 6s ago
```

### Job file (if appropriate)

```hcl
job "example" {
datacenters = ["dc1"]

update {
max_parallel = 1
health_check = "task_states"
auto_promote = false
canary = 1
stagger = "30s"
}

group "cache-0" {
count = 3

task "redis" {
driver = "docker"

config {
image = "redis:3.2"
}

env {
ENV_KEY = "val1"
}

resources {
cpu = 500
memory = 256
}
}
}
}
```

Start with the generic scheduler and reconciler behavior described in the issue, then reproduce the task-group rename using the provided Nomad commands and job file. Determine how the update stanza and max_parallel are expected to apply when a group is removed and inserted. Done should include an agreed behavior, with documentation or implementation and tests covering the rename scenario.

Written by the indexing model from the issue text.