Here are a few posts I can find on the issue mentioned above (though the summary is accurate: we are unlikely to get such information and can only trust staff that this information is not generally actually helpful):
For this specific issue, you may have seen some tips you can use to make Dashboards load faster. If you haven't, here's a summary:
- Instead of "Use all devices," select only the specific devices you are using on that Dashboard
- Make sure to remove or fix any "ghost" tiles (ones that point to a device that no longer exists, displaying something like "unknown")
HubConnect (or Link to Hub + Hub Link) allow you to "share" devices from one hub to another. You have to choose the specific device(s), and then on the other hub it's just another device you can use to automate (run commands on, read attributes from, etc.). People have different reasons for using multiple hubs, but your use case--or at least what I assume you are implying would be yours--is one: running custom code on one hub in case it runs awry, while keeping the other hub with likely "good" apps (e.g., stock apps) and some or all "real" (Z-Wave and Zigbee) devices, depending on how you wan to divide the rest of that up. Some people use multiple hubs for other reasons, e.g., because they have a ton of Z-Wave devices and perceive a performance benefit from splitting up the limited I/O of this protocol. Others segregate known "problem devices" to another hub, like most Zigbee bulbs. Some do a combination of these.
So to directly answer this question, nothing is inherently made available except that which you make available; you can do everything (all devices plus location mode) if you want, but I'd personally just choose the actual devices I need. If you keep one "coordinator" hub for Dashboards, Alexa, and whatnot, then you'd certainly need that hub to have any relevant devices, which I suppose might be almost everything regardless. 
