Updated the URDFPrefabMaker to avoid referencing SDF link names

[lemonade-dm]

Link names within an SDF are only unique with a tag, not unique across the entire SDF document.

The CreatePrefabTemplateFromUrdfOrSdf function has been modified to create a JointMapper object which stores a mapping of joints to all their parent and child links. It also creates a mapping of joints to the model they are directly attached to.

A similar LinkMapper structure has been added for SDF links, which stores a mapping of the link to their attached model.

Instead of CreatePrefabTemplateFromUrdfOrSdf creating a mapping of SDF link name to O3DE Entity, that mapping has now been updated toi map the SDF link pointer to O3DE Entity. This ensures that SDF links with same name are not lost when mapping to O3DE entities. This is needed as SDF links and joints are only required to be unique relative to their attached model. Two links on different models, can have the same link name (http://sdformat.org/tutorials?tut=composition_proposal#1-6-proposed-parsing-stages)

Updated the Robot Importer Visitor callback to require "ModelStack" parameters, which passes in the stack of sdf::Model objects that were visited on the way to the sdf element being visited be that a link, joint or nested model. Also updated the GetAllJoints and GetAllLinks function to store a fully qualified name for the joint/link including the ancestor model names when returning the name to joint and name to link map respectively.

[michalpelka]

I've rebased it locally to test but I've one test failing:

143: [  FAILED  ] UrdfParserTest.TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly

Here is rebased branch : https://github.com/RobotecAI/o3de-extras/tree/mp/sdf-multi-model-support-fixes_rb
How to test PR? I've tested it against the URDF, but I think better would be to run it against multimodel SDFs. Is there any test files?
Edit:
The test TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly fails for development as well 🙈

[lemonade-dm]

I've rebased it locally to test but I've one test failing:

143: [  FAILED  ] UrdfParserTest.TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly

Here is rebased branch : https://github.com/RobotecAI/o3de-extras/tree/mp/sdf-multi-model-support-fixes_rb How to test PR? I've tested it against the URDF, but I think better would be to run it against multimodel SDFs. Is there any test files? Edit: The test TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly fails for development as well 🙈

That test failure isn't due to this change, but do to the combination that the UrdfParserTests.cpp is using an updated version of the AZ::IO::Path CompareRelative internal function to fix comparing paths with paths in them added in this commit https://github.com/o3de/o3de/blame/3d70b5c813119f14723409dd01b339a6be8afded/Code/Framework/AzCore/AzCore/IO/Path/PathParser.inl#L773-L776.
That commit is only available in the o3de development not stabilization/2310 currently, so the test only pass in when using the o3de repo development branch
The test passes in if using o3de development with the o3de-extras repo.

[lemonade-dm]

I've rebased it locally to test but I've one test failing:

143: [  FAILED  ] UrdfParserTest.TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly

Here is rebased branch : https://github.com/RobotecAI/o3de-extras/tree/mp/sdf-multi-model-support-fixes_rb How to test PR? I've tested it against the URDF, but I think better would be to run it against multimodel SDFs. Is there any test files? Edit: The test TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly fails for development as well 🙈

That test failure isn't due to this change, but do combination that the UrdfParserTests.cpp is using an updated version of the AZ::IO::Path CompareRelative internal function to fix comparing paths with paths in them added in this commit https://github.com/o3de/o3de/blame/3d70b5c813119f14723409dd01b339a6be8afded/Code/Framework/AzCore/AzCore/IO/Path/PathParser.inl#L773-L776. That commit is only available in the o3de development not stabilization/2310 currently, so the test only pass in when using the o3de repo development branch The test passes in o3de development

Now there are three choices here that can resolve or mitigate the issue.

1. As the failure is only in the UrdfParserTest.TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly Unit Test and not in any actual functionality due to that test comparing a path of

"/home/foo/ros_ws/install/foo_robot/meshes/bar.dae" == "/home/foo/ros_ws/install/foo_robot/./meshes/bar.dae"

The test failure can be ignored.

2. The AZ IO Path changes can be cherry picked into stabilization/2310 and that would resolve the test failure
3. The test code can be modified to use a call to LexicallyNormal() before the comparison, which would have the comparison succeed as in an older revision of the code

[lemonade-dm]

Here is rebased branch : https://github.com/RobotecAI/o3de-extras/tree/mp/sdf-multi-model-support-fixes_rb How to test PR? I've tested it against the URDF, but I think better would be to run it against multimodel SDFs. Is there any test files?

For test files, the aws-robomaker-hospital-world repo hospital.world SDF file and the aws-robomaker-bookstore-world repo bookstore.world contains a tag with multiple models within them and those files are used for testing this change.

[lemonade-dm]

I've rebased it locally to test but I've one test failing:

143: [  FAILED  ] UrdfParserTest.TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly

Here is rebased branch : https://github.com/RobotecAI/o3de-extras/tree/mp/sdf-multi-model-support-fixes_rb How to test PR? I've tested it against the URDF, but I think better would be to run it against multimodel SDFs. Is there any test files? Edit: The test TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly fails for development as well 🙈

That test failure isn't due to this change, but do combination that the UrdfParserTests.cpp is using an updated version of the AZ::IO::Path CompareRelative internal function to fix comparing paths with paths in them added in this commit https://github.com/o3de/o3de/blame/3d70b5c813119f14723409dd01b339a6be8afded/Code/Framework/AzCore/AzCore/IO/Path/PathParser.inl#L773-L776. That commit is only available in the o3de development not stabilization/2310 currently, so the test only pass in when using the o3de repo development branch The test passes in o3de development

Now there are three choices here that can resolve or mitigate the issue.

1. As the failure is only in the `UrdfParserTest.TestPathResolve_ValidPathRelativeToAncestorPath_ResolvesCorrectly` Unit Test and not in any actual functionality due to that test comparing a path of

"/home/foo/ros_ws/install/foo_robot/meshes/bar.dae" == "/home/foo/ros_ws/install/foo_robot/./meshes/bar.dae"

The test failure can be ignored.

2. The AZ IO Path [changes](https://github.com/o3de/o3de/commit/183d594453d712e4d9991f96e502bd0f3e847faa) can be cherry picked into stabilization/2310 and that would resolve the test failure

3. The test code can be modified to use a call to `LexicallyNormal()` before the comparison, which would have the comparison succeed as in an [older revision of the code](https://github.com/o3de/o3de-extras/pull/505/commits/3daf84ba43ac19ed0a37e68bdea47fe8f84cce3b#diff-51028f9e7742ab6c69b09457939ea407a0b1ff06edb5d451eaf5e8c0e057db36L982-L985)

The AZ::IO::Path changes have been cherry picked to the o3de repo stabilization/2310 branch in PR o3de/o3de#16798, so the Unit Test will now succeed when using the stabilization branch

[michalpelka]

My previous concerns were unrelated to this change. Approving.