Move children schema to strands remove use of deprecated module

Contents (#112)

Operations

• Add ReadTheDocs config file
• Fix codecov upload
• Use reusable workflows
• Test with python3.9 too
• Upgrade actions

Dependencies

• Fix documentation requirements

Refactoring

• Move children and manifest schemas to Strands

Fixes

• Fix accordion usage in docs
• Swap deprecated/removed pkg_resources for importlib

Use trusted publisher for PyPI

Contents (#110)

Operations

• Update workflows to use PyPI trusted publisher and run tests before release

Chores

• Update pre-commit config

Enable code scanning

Summary

Enable code security scanning and update CI workflows.

Contents (#109)

Operations

• Add codeql.yml workflow to enable code-scanning
• Use new octue/check-semantic-version and octue/generate-pull-request-description GitHub actions
• Update versions of other GitHub actions used in workflows

Chores

• Upgrade flake8 pre-commit hook

Update manifest schema

Contents (#108)

Fixes

• Update manifest schema to be compatible with files fields in serialised datasets and to remove outdated fields within serialised files

Remove deprecated code

Summary

Contents (#106)

IMPORTANT: There is 1 breaking change.

Refactoring

• 💥 BREAKING CHANGE: Removes the deprecated code that allowed datasets to still be provided as lists in manifests and manifest schemas.

Revert "Add output location strand"

Summary

Reverts #104.

Contents (#105)

Reversions

• Revert "Add output location strand"

Add output location strand

Summary

Add the ability to provide an output location in twine.json - this is a path to a Google Cloud Storage directory where all output datasets will be uploaded to.

Contents (#104)

New features

• Add output location strand

Allow optional datasets in manifest strands

Contents (#103)

Enhancements

• Allow datasets in manifest strands to be optional

Allow path-only datasets in manifest schema

Summary

Allow a manifest's datasets to be specified as paths, allowing metadata to be gathered dynamically from these datasets instead of hard-coded in a manifest serialisation.

Contents (#101)

IMPORTANT: There are 2 breaking changes.

Enhancements

• BREAKING CHANGE: Make Twine.available_strands a set instead of a tuple
• Update manifest schema to allow path-only datasets
• Raise error if a dataset specified in a manifest strand of the twine is missing from the corresponding manifest
• Add Twine.available_manifest_strands property

Refactoring

• BREAKING CHANGE: Move dataset files tag checking into octue-sdk-python at octue.runner.Runner._validate_dataset_file_tags

Refactoring

• Move manifest and twine migrations into migrations subpackage

Use latest JSON metaschema

Summary

Use the latest JSON metaschema and switch to HTTPS to access it.

Contents (#100)

Enhancements

• Use latest JSON metaschema 2020-12 and access it via HTTPS

Fixes

• Fix labels ref in manifest schema

Dependencies

• Upgrade to jsonschema==4.4.0