feat(datasets) Add a check for same object partitioners

[adam-narozniak]

Issue

The partitioners specification as dict of split to the partitioner object should not have the same objects (value of the dict) assigned to different keys (even if the same type of partitioning happens for different splits).

Proposal

This PR adds a check to ensure this correctness (without it, the code "fails" silently, the non-first used split will not be partitioned but the partitions of the first divided split will be returned).

Examples

This would fail silently

iid = IidPartitioner(num_partitions=10)
fds = FederatedDatasets("mnist", partitioners={"train": iid, "test": iid})

With this PR, it would raise an error.

[jafermarq]

Suggested change

	Check if the multiple partitioner objects are not the same Python object, which
	is not allowed, as the partitioner objects should be independent (one
	partitioner per split).
	Check if each partitioner is a different Python object. Using the same partitioner
	for different splits is not allowed.

[jafermarq]

I think this is clearer