introduce an opinionatedly better statsd helper that gives you access…

[raginpirate]

Whats the diff

This opinionatedly removes the yielded block from statsd_count_sucess and makes developers use tags instead.
The tags proc now supports a result parameter to let developers tag data from any datapoint they would have available to them if they were writing their instrumentation inline (such as a tag like result: T.any(:success, :failure, :error), which is impossible using statsd helpers today).

Note: the name right now is a joke, we can fix the name later 😂

Example of this in use

class Q
    extend(StatsD::Instrument)

    attr_accessor :owner_type
    def initialize(owner_type: "shop_pay_buyer")
    	@owner_type = owner_type
    end

	def y(random_argument:, other_random_argument:)
		"hello"
	end

	statsd_count_but_better(:y, "tester", tags: ->(obj, args, result) { puts obj.owner_type; puts args; puts result; })
end

Q.new.y(random_argument: 3, other_random_argument: "so much clarity!")

> shop_pay_buyer
> {random_argument: 3, other_random_argument: "so much clarity!"}
> hello

✅ What

🤔 Why

👩🔬 How to validate

Checklist

• I documented the changes in the CHANGELOG file.

[raginpirate]

I need to fix this, probably requires a bool use_result = false for real nilable result objects.
It should also accept errors that it can hand over to the tag proc too.