vagrantfile: expose Nomad and Consul APIs to local machine.

* debug: default node-id to all
* debug: align cli help and website documentation

Log the failure error when the agent fails to start. Previously, the
agent startup failure error would be emitted to the command UI but not
logged. So it doesn't get emitted to syslog or `log_file` if they are
set, and it makes debugging much harder. Also, logging the error again
before exit makes the error more visible: previously, the operator
needed to scroll to the top to find the error.

On a sample failure, the output will look like:
```
==> WARNING: Bootstrap mode enabled! Potentially unsafe operation.
==> Loaded configuration from sample-configs/config-bad
==> Starting Nomad agent...
==> Error starting agent: setting up server node ID failed: mkdir /path-without-permission: read-only file system
    2021-10-20T14:38:51.179-0400 [WARN]  agent.plugin_loader: skipping external plugins since plugin_dir doesn't exist: plugin_dir=/path-without-permission/plugins
    2021-10-20T14:38:51.181-0400 [DEBUG] agent.plugin_loader.docker:...

Closes: #11395 .

The system scheduler should leave allocs on draining nodes as-is, but
stop node stop allocs on nodes that are no longer part of the job
datacenters.

Previously, the scheduler did not make the distinction and left system
job allocs intact if they are already running.

I've added a failing test first, which you can see in https://app.circleci.com/jobs/github/hashicorp/nomad/179661 .

Fixes https://github.com/hashicorp/nomad/issues/11373

Pick up https://github.com/golang/snappy/pull/56 to handle arm64 architectures to fix panics. tldr; Golang 1.16 changed `memmove` implementation for arm64 requiring additional cpu registers that snappy wasn't preserving in its assembly implementation.

Other projects have experienced this issue as well, searching for `encode_arm64.s:666` on your favorite search engine will reveal some.  Vault updated the dependency earlier this August: https://github.com/hashicorp/vault/pull/12371 .

I believe this issue affects Nomad 1.2.x and 1.1.x. Nomad 1.0.x use Golang 1.15 and isn't affected. However, backporting the change to 1.0.x should be harmless.

Fixed https://github.com/hashicorp/nomad/issues/11385 .

* Update Consul Connect to Consul service mesh
* Apply suggestions from code review

* Impl Fathom analytics

* Actually install fathom-client

* Use analytics package instead of direct impl

* Remove explicit fathom-client dep

* Upgrade platform analytics package

website: fixup link formatting within interpolation doc.

Update HashiConf alert-banner expiration

Updates the HashiConf Alert Banner expiration to 10/20 @ 11pm (PT)

Add support for --init to docker driver.

Code cleanup: Remove extra if clause.

Fix a bug where the scheduler may panic when preemption is enabled. The conditions are a bit complicated:
A job with higher priority that schedule multiple allocations that preempt other multiple allocations on the same node, due to port/network/device assignments.

The cause of the bug is incidental mutation of internal cached data. `RankedNode` computes and cache proposed allocations  in https://github.com/hashicorp/nomad/blob/v1.1.6/scheduler/rank.go#L42-L53 . But scheduler then mutates the list to remove pre-emptable allocs in https://github.com/hashicorp/nomad/blob/v1.1.6/scheduler/rank.go#L293-L294, and  `RemoveAllocs` mutates and sets the tail of cached slice with `nil`s triggering a nil-pointer derefencing case.

I fixed the issue by avoiding the mutation in `RemoveAllocs` - the micro-optimization there doesn't seem necessary.

Fixes https://github.com/hashicorp/nomad/issues/11342 

Signed-off-by: Shishir Mahajan <smahajan@roblox.com>

website: Update alert banner for HashiConf

Final cleanup/closer exp date

Signed-off-by: Shishir Mahajan <smahajan@roblox.com>

Meant for development purposes only, so one can compile binary on a
macos host then start a Docker container or scp the binary to a linux
host easily.

The resulting binary is statically linked and has very subtle
differences. e.g. static binaries use go native network stack that
honor /etc/hosts and /etc/resolve differently from the glibc
implementation. In development environment, I don't expect these to
materially change our experience.