refactor(orderGraphValidator): Isolate OrderGraphValidator and use vda5050_types

[shawnkchan]

Summary

This PR addresses #8. This branch was created from the develop branch.

In this branch:

1. OrderGraphValidator is now isolated from OrderManager
2. OrderGraphValidator uses the hard types in vda5050_types
3. There is no interaction with any state logic
4. Unit tests are passing

[sauk2]

Thanks for the great work! I have provided my first phase of review comments.

I haven't read through the source file properly yet. These comments are mostly on the exposed API.

[sauk2]

Some more comments regarding the functional logic. I feel there are a few validation checks still missing,

• All node sequence IDs should be even and all edge sequence IDs should be odd.
  This follows from the structure of a valid order - an order must contain at least one node (the AGV’s current position), which always starts at sequence ID 0. From there, nodes and edges should alternate.
• The validator should ensure that no two nodes or edges share the same sequence ID. Duplicate sequence IDs would break the ordering during execution.
• Sequence IDs should form a continuous chain with no gaps so there should be a check to search for any missing sequence IDs.
• There should be a clear boundary between base and horizon entities. Once the first unreleased node or edge is encountered, all following nodes and edges must also be unreleased.

[sauk2]

I don't think you need to bring in ament_cmake_gtest separately. The previous ament_cmake_gmock most probably does that

[sauk2]

The current validation logic seems incorrect. Right now, the code compares every edge’s start_node_id only against the first node of the order and end_node_id only against the last node of the order. However, start_node_id and end_node_id on an edge describe the two nodes that the specific edge connects, not the global start and end of the entire order.

For a linear order, the correct relationship is sequence-based, i.e, the edge with sequence ID n should connect the node with sequence ID n-1 (as its start_node_id) and the node with sequence ID n+1 (as its end_node_id)

Because this validation depends entirely on the ordering of sequence IDs, it should not be done using only the first and last node. Instead, the check should be integrated into the part of the validation where you are already confirming that sequence IDs form a proper graph. That way you can directly verify that each edge connects the two adjacent nodes it is supposed to.

[Johnarman1398]

@sauk2 For review,

implements validation for VDA5050 order graphs to ensure basic order and sequencing rules are met before execution.

Features:

• is_valid_graph()

1. Enforces a linear chain where Nodes = Edges + 1.
2. Checks for parity (Nodes=Even, Edges=Odd) and continuity (no gaps in Sequence IDs)
3. Ensures "released" nodes never appear after "unreleased" nodes.
4. Ensure edge's start_node_id and end_node_id are correctly labelled.

• is_valid_order_update()

4. Logic to ensure new Order updates stitch correctly to the last released node of the base graph