Cumulus-library

Breaking changes
* Actions/builders APIs now primarily work from StudyConfig and StudyManifest objects
* Minimum python is now 3.11
* Changed `_code_system` in tables to just `_system`
* Replace `core__medicationrequest` with `core__medication`, removed previous version
* Removed `core__study_period`, offloaded study definitions to downstream studies

Deprecation warning
* `core__observation` is now considered deprecated - use a specific observation profile table, if one exists. Once we have more profile coverage, `core__observation` will be removed.

Major changes
* Added a generic way to log to a study's transaction table
* Allow studies to specify their own unique schema (intended for independent data sets, like code systems)
* Allow studies to specify their own CLI args with `--option`

Minor changes/bugfixes
* Several new generic templates & jinja syntax helpers
* Added StudyManifest, StudyConfig, BaseTableBuilder, and CountsBuilder to the package-level API
* Use `cumulus-fhir-support` for reading ndjson
* Fixed Athena's ability to export "decimal" types
* Updated CLI to use terms like `database` and `schema` in the context of a specific database's nomenclature
* Added $expand to the UMLS VSAC API, and paginate results
* `clean` CLI arg now removes views
* `generate-sql` now masks S3 paths in output

**Full Changelog**: https://github.com/smart-on-fhir/cumulus-library/compare/v2.3.0...v3.0.0

- Duckdb support for different JSON folder formats
- Added import command for re-loading previously exported studies
- Improvements to exporting for very large files
- Added option for studies to have an explicit schema under specific conditions

-Bugfix for some CLI args
-Using manifests, rather than filepaths, for discovering study names
-When tables do not exist, fall back to empty tables so queries dont fail
-updated list of official study CLI targets
-Improvements to partial string matching.

- Added ability to create study tables from static files
- Updated medication tables generation
- This includes several fixes to core_medication and core_medication-request
- Added API library for UMLS API endpoints
- Improved handling of missing patient extensions
- Improved handling of observation lab/vital signs
- Handling of FHIR `component` as a denormalized table
- Improvements to table schema parsing
- Fixed handling of Encounter.type field
- Improved handling of errors in the `upload` command

This hotfix release makes the following changes:

- fixes a typo in the encounter.serviceType_code_* column names
- fixes a case where `cnt_subject_ref` was not being appropriately specified

Breaking changes:
- Table columns renamed to better match FHIR names
- Overhaul of how codings are handled - now dynamically detected and joined as appropriate
- All core tables are strictly 2 dimensional; unnesting should not be required if using core as the basis.
- Minimum python version is now 3.10

Other major changes:
- Core builds :should: now gracefully react to missing nested data from the EHR
- magical string `cumulus_none` added to reflect cases where null data was recieved from a downstream system
- Added missing required/must support fields from US Core R4
- Overhaul of template infrastructure to better centralize/reuse code
- Added duckdb support, which we are now using for unit tests
- Propensity score matching added as an available statistical library

Minor changes:
- Schema validation is now more robustly case insensitive
- Added additional coding fields to the discovery study
- Added `--archive` cli command for storing study data outside athena
- Added `--generate-sql` cli command for making example output from a study's jinja templates
- Added `--generate-md` cli command to create markdown documentation directly from a database
- Switched to ruff for linting/security
- `id` field is now always the first column in a FHIR resource table