No. The way the code is designed, that's explicitly the responsibility of the code that creates the analysis context collection. At the moment, the code that watches the file system lives in the analysis_server package.
There are a few reasons for this.
1. The applications that we were previously aware of that need to create a collection don't want to listen to file changes. This is the first request we've had for such a feature.
2. When changes occur on the file system, the intended behavior is to discard the existing collection and create a new one. Doing so is a fairly light-weight operation, and it reduces the probability of inconsistencies being introduced while attempting to update the collection in place.
3. In the server we need to also take into account the state of files that have been edited but not saved. If unsaved changes exist in a file that's modified the server needs to continue to analyze the code in the edit buffer, not the code on disk, so it doesn't make sense to put that functionality in code that doesn't know about the client.