Add Configuration API

[p-lambert]

Overview

This PR provides the core components that will power the new configuration API. All integrations should include the Base module which in turn includes the Registerable and the Configurable modules.

By doing so, the integration will be globally accessible through the name specified in .register_as method call and all of its configuration parameters can be exposed through a series of .option declarations.

As a result, users will be able to configure everything within a single configuration block such as:

Datadog.configure do |c|
  c.use :rails, default_service_name: 'foobar'
  c.use :aws
  c.use :http, distributed_tracing_enabled: true
  c.use :dalli
end

And the integrations will look like:

module Datadog
  module Contrib
    module HTTP
      include Base
      register_as :http
      option :distributed_tracing_enabled, default: false

      def patch
        # Instrumentation Code
      end
    end
  end
end

These changes should obviate the interaction with Datadog::Monkey and Datadog::Pin in the future.

Meta Options! (For Integration Development)

The .option method provided by Configurable has the following features:

Default Values

A default value can be specified as:

option :foo, default: 'bar'

Notice that we also support lazily evaluated defaults, so you can have something like:

option :foo, default: -> { 1 + 1 }

Custom Setters

You can specify custom setters for your option param using either the setter option or providing a block:

option :shout, setter: ->(value) { value.upcase }
# alternative syntax
option(:shout) { |value| value.upcase }

Dependent Options

module AwesomeIntegration
  include Datadog::Contrib::Base
  option :tracer, default: Datadog.tracer
  option :enabled, depends_on: [:tracer] do |value|
    get_option(:tracer).enabled = value
  end
end

[palazzem]

Perfect design. It allows us to provide configurations like it should be: crystal clear.