S3 browser - need to implement lazy loading

[smalers]

There is a performance hit and potentially higher AWS charges if full bucket lists are shown in the S3 browser. This is particularly the case if the initial display and root node does not correspond to a folder. This could be a bigger issue when a bucket is used for many system files, such as at TriLynx Systems. Lazy loading needs to be implemented as followss:

1. List only top-level folders and files.
2. When a folder is selected to expand, read its contents with a request.
3. It may be OK to just lazy load based on top-level folder selection, as a first implementation.
4. The tree needs to be smart enough to only lazy load when needed.