Extend extended loader to avoid state leakage across asset parses

[j1george]

https://fieldwire.atlassian.net/browse/ME-6870

Resets extended-loader parse state so later asset parses do not reuse stale loaded-state.

This updates the extended loader to clear the loaded-state for buffer/decode work before a new asset parse begins, while still preserving the intended once-per-asset behavior within a single parse. The goal is to prevent later asset loads on the same loader from incorrectly skipping required work because state from an earlier parse was left behind.

This change is limited to the extended-loader state lifetime issue. It does not add Android-side multi-model integration, sample or harness support, or broader federated BIM behavior. It also does not change placement, framing, picking, transforms, or app lifecycle handling.

update: previously was using a set to track load. current implementation just keeps the boolean and resets it on each new asset

Screenshots

N/A

Testing Notes

What did you do to test this PR?

• added regression coverage for repeated parses on the same extended loader
• built and ran test_gltfio; the new repeated-parse test passed
• ran local pre-Android multi-model validation using uncommitted gltf_viewer harness-only changes outside this PR
• confirmed multiple models rendered in one scene in the local viewer harness
• built new filament AARs from this change and validated them through the existing Android POC multi-model path
• confirmed rendering, navigation, and picking worked in the Android POC
• observed no discernable performance regression versus the POC filament changes during this validation

[github-actions]

Please add a release note line to NEW_RELEASE_NOTES.md. If this PR does not warrant a release note, add the 'internal' label to this PR.

[Copilot]

Pull request overview

Fixes a state-lifetime issue in the extended glTF loader where per-parse buffer/decode bookkeeping could leak across multiple asset parses performed by the same AssetLoader, causing later loads to incorrectly skip required work.

Changes:

• Introduce per-parse bookkeeping in AssetLoaderExtended and reset it via a new beginAssetLoad() method.
• Replace the previous single boolean “buffers loaded” flag with a set-based tracker and ensure failed loads don’t poison subsequent retries.
• Invoke beginAssetLoad() from FAssetLoader::createInstancedAsset before starting a fresh parse.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

File	Description
libs/gltfio/src/extended/AssetLoaderExtended.h	Adds beginAssetLoad() and new per-parse tracking container for buffer-load bookkeeping.
libs/gltfio/src/extended/AssetLoaderExtended.cpp	Gates loadCgltfBuffers + meshopt decode behind the new per-parse tracking.
libs/gltfio/src/AssetLoader.cpp	Resets extended-loader per-parse state before creating a new asset from parsed cgltf data.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated no new comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.