WIP first crack at _suppress_initial support

[znewsham]

Would you be interested in a completed PR to allow _suppress_initial support?

The specific use case: I'm counting documents and need reactive updates to the counts, but I don't need to pull all the documents initially.

[theodorDiaconu]

What does it solve? @znewsham

[znewsham]

Using the native oplog __supress_initial allows developers to create an observer without first pulling in the entire dataset covered by the query.

For example, if I wanted to publish a count of documents, I don't need to pull in every document, I just need to run a count, then rerun the count when the dataset changes (or manually modify the count when added/removed events are seen).

Even though __supress_initial is a private property that was never documented, the lack of it feels like a regression and not having it makes some tasks more challenging