Support S3 paths in h2o.load_grid and h2o.loadGrid

[hutch3232]

Is your feature request related to a problem? Please describe.
h2o's python package supports S3 paths in h2o.save_grid yet h2o.load_grid throws a "Not Implemented Error". The same is true for the corresponding R package functions. It seems like you should be able to save/load from the same paths like you can do for non-grid model objects.

Describe the solution you'd like
Implement S3 path support for h2o.load_grid and h2o.loadGrid.

Describe alternatives you've considered
Manually copy the grid files to a temporary file storage then apply h2o.load_grid.

[hutch3232]

I don't know java, but looking around a bit, it seems possibly the solution is to implement the getParent method for S3 paths.

h2o-3/h2o-core/src/main/java/hex/grid/Grid.java

Lines 643 to 644 in facf8f4

	final Persist persist = H2O.getPM().getPersistForURI(gridUri);
	final String gridDirectory = persist.getParent(gridUri.toString());

h2o-3/h2o-core/src/main/java/water/persist/Persist.java

Lines 226 to 228 in facf8f4

	public String getParent(String path) {
	throw new RuntimeException("Not implemented");
	}

https://github.com/h2oai/h2o-3/blob/master/h2o-persist-s3/src/main/java/water/persist/PersistS3.java
does not contain an override.

S3 doesn't really have "directories" but I think it would work to just use "/" to strip off the end.