- Update the AWS provider.
- Use the common tags option in the provider and improve tagging usage.
- Address a few warnings from the updated provider.

Repeatable builds, except for variations in dependencies.

Mostly ignore things I'm not interested in but it did find somethings.

It turns out that Twilio is quite expensive, more than I willing to pay
for this service. So I'm going with MessageBird.

AWS' SMS delivery has been unreliable to say the least. Replace it with
Twilio but that's more complex:

- A new Lambda function to send SNS messages as SMS messages with
  Twilio.
- Add the function as a subscription to the SNS topic.

Docstrings for the different checks.

For cases like moving to a different region.

Also, with the newer Pylint version, address some new issues.

SMS sending fails ALL OF THE  TIME. Try a different region that may work
better with my israeli phone number.

it's the default before_script action, but it would only affect the
pre-commit job (all of the others have their own before_script). It
would allow me to not skip the terraform pre-commit checks.

The pre-commit template now runs in .pre. Reorder the other jobs.

- Check base class, all checks derive from it and have a common calling
convention.
- Add retries.
- Add a DNS zone transfer check. Turns out if fails on ns1. WIP.
- A wrapper around checking web sites. Just provide a list of URLs.

Outbound traffic on port 25 is restricted by AWS. Removing the
restriction will raise the cost of the project.

The bucket already has the environment in it.

Ease debugging.

There's a limit of 5 targets per rule. So rule per function it is.

The triggers will be created last (after the rule, the functions and the
aliases are created). Without them, there won't be any invocations and
obviously the alarms will trigger.