Octue

<!--- SKIP AUTOGENERATED NOTES --->
Contents

Enhancements
- [x] Let `FilterContainer.one` be used with no filters on empty containers

<!--- END AUTOGENERATED NOTES --->

<!--- START AUTOGENERATED NOTES --->
Contents

Fixes
- [x] Upload dataset metadata file for cloud directory datasets when getting a dataset from a cloud directory with no existing metadata file

Refactoring
- [x] Rename `dataset.json file` to `dataset_metadata.json`
- [x] Simplify `Dataset.from_cloud`

<!--- END AUTOGENERATED NOTES --->

<!--- SKIP AUTOGENERATED NOTES --->
Contents

Operations
- [x] Update pull request workflow

Refactoring
- [x] Use more resilient method to list blobs in a cloud directory

<!--- END AUTOGENERATED NOTES --->

<!--- START AUTOGENERATED NOTES --->
Contents

Fixes
- [x] Allow `Manifest` construction from any serialized cloud dataset

<!--- END AUTOGENERATED NOTES --->

<!--- SKIP AUTOGENERATED NOTES --->
Contents

Enhancements
- [x] Give cloud `scandir` option to include subdirectories

Fixes
- [x] Ensure `Dataset` construction from a cloud directory with no `dataset.json` file does not include subdirectories

Dependencies
- [x] Upgrade GCP storage emulator version

Operations
- [x] Unify GitHub workflow skip commands

<!--- END AUTOGENERATED NOTES --->

Summary
Allow simpler instantiation of datasets from the cloud.

<!--- START AUTOGENERATED NOTES --->
Contents

Enhancements
- [x] Allow instantiation of datasets from any cloud directory rather than just those that have a dataset serialised as a `dataset.json` file in them

<!--- END AUTOGENERATED NOTES --->