Formalize specification/documentation of how transforms interact with heuristics and queries

[jleibs]

There are several open questions at the moment regarding where transforms fit into the heuristic system.

• Do they only impact "Spatial" views or can they be used to split content in different kinds of views.
• Is there a mechanism to "query" entities up to the next transform?
• Is there actually a difference between a "DisconnectedSpace" and a Null "SpatialTransform?"
• Can a pinhole transform be converted to a rigid transform or does it create a permanent topological disconnect between 2d and 3d spaces.
• Can 3D content be added to a 2D space without a transform?

This would improve clarity for users in issues like:

• DisconnectSpace and TimeSeriesScalar #4465

[Wumpf]

We got answers for all of these by now and most of them even coded up:

• Do they only impact "Spatial" views or can they be used to split content in different kinds of views.
  • only impact spatial views. SpaceInfo should go away
• Is there a mechanism to "query" entities up to the next transform?
  • transforms propagation is part of general property propagation
• Is there actually a difference between a "DisconnectedSpace" and a Null "SpatialTransform?"
  • Yes. A disconnected space changes the spacial topology. A null spatial transform is handled as identity
• Can a pinhole transform be converted to a rigid transform or does it create a permanent topological disconnect between 2d and 3d spaces.
  • no. It's a permanent topological disconnect
• Can 3D content be added to a 2D space without a transform?
  • no

@jleibs plz check over these and close if we're on the same page. If not then this ticket is clearly not done 😄

[jleibs]

Agree we're on the same page. Would be great to capture this somewhere durable. Almost like we need a spatial-transform architecture markdown doc. (Which is where we would also add future notes on semantics for things like time-varying transforms #1522)