Datastore: Allow retry and timeout configuration on a per instance level

[Subaku]

Seems the configuration for these two items are buried in the code and can't be configured on a per use basis. When debugging connectivity issues the default timeout takes forever and would love to set a shorter timeout on my calls to query.fetch(...) for example.

Or perhaps I've missed something in the docs?

[tseaver]

Sketch of implementation:

• Add retry and timeout args to Batch.commit and plumb through to the underlying RPC in _commit.
• Add retry and timeout args to Client.get, and plumb through to the underlying RPC.
• Add retry and timeout args to Client.get_multi, and plumb through to the underlying RPC.
• Add retry and timeout args to Client.put, and plumb through to the underlying RPC (Batch.commit).
• Add retry and timeout args to Client.put_multi, and plumb through to the underlying RPC (Batch.commit).
• Add retry and timeout args to Client.delete, and plumb through to the underlying RPC (Batch.commit).
• Add retry and timeout args to Client.delete_multi, and plumb through to the underlying RPC (Batch.commit).
• Add retry and timeout args to Client.allocate_ids, and plumb through to the underlying RPC.
• Add retry and timeout args to Client.reserve_ids, and plumb through to the underlying RPC.
• Add retry and timeout args to Transaction.begin, and plumb through to the underlying RPC.
• Add retry and timeout args to Transaction.rollback, and plumb through to the underlying RPC.
• Add retry and timeout args to Transaction.commit, and plumb through to the base class method.
• Add retry and timeout args to query.Iterator ctor, and plumb through to the underlying RPC in _next_page.
• Add retry and timeout args to Query.fetch, and plumb through to the iterator ctor.