JIT: annotate pseudo-edges in pgo schema and/or internal model

[AndyAyersMS]

Sparse edge instrumentation will sometimes add pseudo-edges into the flow graph to provide a complete picture of flow. For instance, there are pseudo-edges from each return back to method start, and from finally exits back to the first block of the finally.

When reconstructing counts from the sparse edge instrumentation data, we need to match up schema flow edges with actual flow edges so we can propagate the counts. Ideally we'd be looking at the exact same flow graph and use the same edge enumeration technique, so we can just internally note which edges were pseudo-edges and so won't have an actual edge, but we might also want to annotate the schema entries so an external client could potentially do the same sort of reconstruction, or so we can sanity check that we have the same flow graph during reconstruction that we had during instrumentation.

[ghost]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch, @kunalspathak
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Sparse edge instrumentation will sometimes add pseudo-edges into the flow graph to provide a complete picture of flow. For instance, there are pseudo-edges from each return back to method start, and from finally exits back to the first block of the finally.

When reconstructing counts from the sparse edge instrumentation data, we need to match up schema flow edges with actual flow edges so we can propagate the counts. Ideally we'd be looking at the exact same flow graph and use the same edge enumeration technique, so we can just internally note which edges were pseudo-edges and so won't have an actual edge, but we might also want to annotate the schema entries so an external client could potentially do the same sort of reconstruction, or so we can sanity check that we have the same flow graph during reconstruction that we had during instrumentation.

Author:	AndyAyersMS
Assignees:	-
Labels:	area-CodeGen-coreclr
Milestone:	-

[AndyAyersMS]

Pseudo-edges are now identified in the internal model via #81738. They are not externalized (persisted to the schema) but not clear if that's important yet.

[AndyAyersMS]

Not clear we'll need this in 8.0 so moving to future.