feat: flat-layout CVAT campaign tools and resilient shard writing

[cau-git]

Summary

Flat-layout CVAT campaign tools

Three new scripts for flat delivery directories (complement to the existing submission-style tools):

Script	Purpose
convert_flat_cvat_deliveries_to_docling.py	Batch-converts flat CVAT XML deliveries → DoclingDocument JSON + HTML; supports parallelism, dry-run, and structured run reports
flat_cvat_deliveries_to_hf.py	Aggregates flat-batch exports into a HuggingFace-ready parquet dataset
cvat_task_batch_uploader_flat.py	Uploads <task>/images/ + <task>/annotations.xml layouts as tasks to a CVAT project

save_shard_to_disk rework

• Returns SaveShardResult (written/skipped counts, next shard id) instead of None
• Automatically splits batches into multiple shards when estimated size exceeds 1.5 GB
• Per-shard ArrowCapacityError recovery with skipped_records.jsonl audit log

Extensions

• cvat_create_annotation_tasks_from_folders.py: added --start-index, --end-index, --chunk-number-offset for resumable chunked runs

[github-actions]

✅ DCO Check Passed

Thanks @cau-git, all your commits are properly signed off. 🎉

[mergify]

Merge Protections

Your pull request matches the following merge protections and will not be merged until they are valid.

🟢 Enforce conventional commit

Wonderful, this rule succeeded.

Make sure that we follow https://www.conventionalcommits.org/en/v1.0.0/

• title ~= ^(fix|feat|docs|style|refactor|perf|test|build|ci|chore|revert)(?:\(.+\))?(!)?: