[Lens] Request managing layer

[flash1293]

We currently have the potential to have many (redundant) requests in flight at once: index pattern fetch, emptiness data fetch, etc. We should write a service which manages the prevention of this (e.g. if you make a N requests to '/foo' while the first is still out, it aborts the first, waits for the abort, then makes the Nth request).

[elasticmachine]

Pinging @elastic/kibana-app (Team:KibanaApp)

[stratoula]

Thank you for contributing to this issue, however, we are closing this issue due to inactivity as part of a backlog grooming effort. If you believe this feature/bug should still be considered, please reopen with a comment.