Closing as out of date.

In Search 5, we build a full model of how each indexed type should be processed while indexing, including any processing of "indexed embedded" types. But we don't do the same for "contained in" processing: for each contained type, we just build the model of all "directly contained" types with includePaths/maxDepth information. And then we resolve the full model at runtime: hoping from one object to another, we detect its type, get the "directly contained" model, and recurse.

That changes in Search 6. We now build a full processing tree for each indexed type, and also a full reindexing resolver tree for each contained type (including navigation through other types and their own "contained" types).

As a result, the algorithms are rather different and were rewritten from scratch: we don't need a "ContainedInRecursionContext" anymore at runtime, since we have a model that includes all the information and don't need to perform any complex resolution at runtime anymore. Thus this ticket doesn't make sense anymore.