Visualize refactor follow up #12930

The refactor introduced a significant amount of changes.

Keep track here of internal changes/feedback to either tidy up or improve the code-base.

• AGREED time-picker auto-refresh should not be implemented in multiple locations (cf. Auto-refresh does not work on dashboard and visualize #12811 (comment)).
• consider trimming the general purpose update_status. The general-purpose serialization adds overhead. Especially for data-refreshes it's an overhead we do not need to incur. We have full-control over the backend and know when the requesthandler is called or not (and thus whether the data is refreshed). This is also buggy, as it flagged data changes where non happened (see Coordinates Map visualization re-creating geohashLayer on map move #12919).
• UNDER CONSIDERATION consider removing responseHandler. This abstraction does not add immediate benefit. Individual visualizations could just make a function call in the render-method to make this transformation individually. Should not overdesign right of the bat, we can introduce this later.
• POC new response handlers for vislib #12661. ES data should be converted to something that can be consumed by majority visualizations (table, series) with no coupling to ES (see response handler PR)
• move shouldQuery logic out of the requestHandler so requestHandlers can be stateless. It would also avoid the bleeding of the vis.reload flag into the requestHandler.
• OK FOR RENAME rename appState.vis to appState.visState. (what is the reason we are storing this on the appState?)
• AGREED consolidate all vis-code in single directory: Consolidate all visualize code in single directory #12675
• some options apply now without having to press play (e.g. tagcloud options, show legend in gauges). Do we want this? It prevents us from using the cancel functionality.
• editormode should be true when designing a visualization (e.g. with sidebar editor), but false on a dashboard. this doesn't seem to be the case.

cc @ppisljar