[gremlin-dotnet] Handle non-bulked results in remote traversals

[xiazcy]

The gremlin-dotnet remote traversal path (DriverRemoteConnection / TransactionRemoteConnection) assumed every server result was a Traverser. It submitted as SubmitAsync<Traverser>, so ResultSet<Traverser> performed an unconditional (Traverser)item cast during iteration.

When a compatible HTTP endpoint returns non-bulked results, e.g. a raw Int64 for g.V().count(), or any traversal run with bulkResults=false, the values arrive as raw scalars rather than Traverser instances, and the cast throws InvalidCastException. In effect, bulkResults=false was unusable on the remote traversal path.

Fix

Submit as SubmitAsync<object> and adapt the stream at the iteration boundary: pass Traverser instances through unchanged, and wrap raw values in new Traverser(item, 1). This mirrors the other GLVs. The generic ResultSet<T> and the plain client.SubmitAsync<T> path are unchanged.

Testing

• Unit tests for the adaptation: raw scalar, bulked Traverser expansion, and a mixed stream (DriverRemoteConnectionTests).
• Remote-traversal integration tests with bulkResults=false for a single scalar (count) and multi-result (values) case (GraphTraversalTests).
• Client-path integration tests confirming bulkResults true/false behavior (GremlinClientTests), matching the Go driver's coverage.

---

VOTE +1