Unified Aero Recipe: Fix for manifest-based path without augmentation

[peterdsharpe]

PhysicsNeMo Pull Request

Description

Stack on top of #1753, #1754

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.
• The CHANGELOG.md is up to date with these changes.
• An issue is linked to this pull request.
• If I am implementing a new model or modifying any existing model, I have followed the Models Implementation Coding Standards.

Dependencies

Review Process

All PRs are reviewed by the PhysicsNeMo team before merging.

Depending on which files are changed, GitHub may automatically assign a maintainer for review.

We are also testing AI-based code review tools (e.g., Greptile), which may add automated comments with a confidence score.
This score reflects the AI’s assessment of merge readiness and is not a qualitative judgment of your work, nor is
it an indication that the PR will be accepted / rejected.

AI-generated feedback should be reviewed critically for usefulness.
You are not required to respond to every AI comment, but they are intended to help both authors and reviewers.
Please react to Greptile comments with 👍 or 👎 to provide feedback on their accuracy.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[melo-gonzo]

I had a look over the changes in dataset.py - all LGTM! Thanks @peterdsharpe

[greptile-apps]

Greptile Summary

This PR fixes a bug in manifest mode where validation data was being passed through the augmented transform chain when augment=True, unlike directory mode which always builds its val dataset with augment=False. The fix introduces _build_manifest_val_dataset and wires it into build_dataloaders so manifest-mode validation now matches directory-mode behavior.

• Adds _build_manifest_val_dataset: returns None (val shares train dataset) when augment=False, or a fresh un-augmented dataset over the same directory when augment=True, relying on the same deterministic reader glob order.
• Updates build_dataloaders to initialise manifest_val_dataset = None before the loop and use it as val_dataset in the manifest branch, with the existing multi-dataset limitation note updated to cover the new variable.
• Adds TestManifestValDataset with two tests that lock down both branches, including a guard assertion that the train dataset actually carries stochastic transforms so the no-augmentation check on the val dataset is non-vacuous.

Important Files Changed

Filename	Overview
examples/cfd/external_aerodynamics/unified_external_aero_recipe/src/datasets.py	Adds _build_manifest_val_dataset helper and wires it into build_dataloaders so validation in manifest mode is always un-augmented when augment=True, matching directory-mode behavior. Logic, variable initialization, and existing NOTE comments around the multi-dataset limitation are all updated consistently.
examples/cfd/external_aerodynamics/unified_external_aero_recipe/tests/test_manifest.py	Adds TestManifestValDataset with two targeted tests: one confirming augment=False returns the None sentinel, and one verifying that augment=True produces a separate dataset object with no stochastic transforms but the deterministic CenterMesh still present. The guard assertion prevents a vacuous pass.

_{Reviews (2): Last reviewed commit: "Merge branch 'main' into psharpe/unified..." | Re-trigger Greptile}

[peterdsharpe]

@greptileai

[coreyjadams]

Looks reasonable to me!