Split out interfaces for Retrier vs HTTP Client

[rShetty]

We have a common big interface for retrier related methods vs those for making http requests.
We could have Client interface for HTTP related methods and Retriable interface for retrier related methods, segregating the interfaces

[darshanime]

(hello, I am learning Go and would like to contribute (please pardon my asking dumb questions!))

Currently we have separate interfaces: Client and Retriable:

// client.go
type Client interface {
	Get(url string, headers http.Header) (*http.Response, error)
	Post(url string, body io.Reader, headers http.Header) (*http.Response, error)
	Put(url string, body io.Reader, headers http.Header) (*http.Response, error)
	Patch(url string, body io.Reader, headers http.Header) (*http.Response, error)
	Delete(url string, headers http.Header) (*http.Response, error)
	Do(req *http.Request) (*http.Response, error)

	SetRetryCount(count int)
	SetRetrier(retrier Retriable)
}

// retry.go
type Retriable interface {
	NextInterval(retry int) time.Duration
}

What does this issue aim to achieve?

On a side note, we could look at merging the Backoff and Retriable interface. They are both related and this could simplify the heimdall API.

type Backoff interface {
	Next(retry int) time.Duration
}

[rShetty]

@darshanime It is always good to have one method interfaces. These help provide composeability when using these/implementing them.

The point of this issue was that we have Client interface which has grown quite a lot and needs splitting. The way we do it still being discussed and debated.

[rShetty]

With #33 this might not be required.