Fix/invalidate on remove

[willscott]

118 introduced the concept of handle invalidation, such that a handle involved in a rename would become stale if attempted to be used after the rename.

This extends that semantic to be also true on deletion.

[willscott]

• I've updated the test to perform and validate renaming.
• I also updated the test to demonstrate directory listings of more than the 1024 sized file handle cache of the caching handler

[willscott]

@DUOLabs333 - can you check if this set of updates works for your use case?

in particular, for the test to work I believe the correct expectation is to invalidate the handle of the removed file, rather than that of the containing directory as we had before. The client i'm using in the test was looking at the mtime of the directory to recognize that there's been a change at that level, which seems to indicate an expectation that the containing directory handle does remain valid across renames within it.

[DUOLabs333]

It seems fine to me.