useSWR onError and onErrorRetry changed behaviour in 1.1.0

[GabrielNastase]

Hello, and thanks for the effort to build this great library.

We have been observed the functionality of onError and onErrorRetry callbacks has changed in the minor 1.1.0

We have cases in our apps where we call useSWR with the same cache key for multiple times, sometimes just for reading data, other times for reading data and handling error scenarios.

Consider this simplified code

const sameFetcher = () => Promise.reject('Ooups');

const ComponentOne = () => {
  const { data } = useSWR('same-key', sameFetcher, {
    onError: (error) => {
      console.log('ComponentOne onError fired', error)
    },
    onErrorRetry: (error, _key, _config) => {
      console.log('ComponentOne onErrorRetry fired', error, _key, _config)
    }
  })

  return <pre>{data}</pre>
}

const ComponentTwo = () => {
  const { data } = useSWR('same-key', sameFetcher, {
    onError: (error) => {
      console.log('ComponentTwo onError fired', error)
    },
    onErrorRetry: (error, _key, _config) => {
      console.log('ComponentTwo onErrorRetry fired', error, _key, _config)
    }
  })

  return <pre>{data}</pre>
}

const App = () => {

  return <div>
    <ComponentOne />
    <ComponentTwo />
  </div>
}

Before 1.1.0 ( before this commit specifically in 1.1.0-beta.4 ) , we could see that both callbacks of the both components were fired. And that is what we expected. In 1.1.0 and after we can see that just the first rendered hook will call back the onError or onErrorRetry

As we have an app with multiple independent panels in the same page, and we need to handle the onErrorRetry different from panel to panels ( often we don't need to handle at all in the first rendered hook ) how are we suppose to do it going forward?

Would be nice if it would have been the option to opt-in the behaviour pre 1.1.0