BallotStyles don't drive the contests.

Fixes BallotLab #82.

Fixes BallotLab #83.

Fixes BallotLab #86.

Fixes BallotLab #84.

Updates #84.

Fixes BallotLab #85.

This isn't quite right: fails if a slate has candidates from different parties.
Deferring on that case for now.

Mark write-in status as boolean (matching the contents of the EDF).
No issue attached.

The error was introduced in the process of folding write-ins into contests.
- Remove convoluted extra step that makes it harder to do this correctly.
- Candidate names are grouped in arrays. Now do the same for parties.
- Use 'is_write_in' for the write-in boolean field.

Handles case where candidates in a slate have different parties.

…entical.

…ssing.

Fixes BallotLab #86.

Significant update of contests:
- Include '@id's of ballot measure contests (candidates already happening).
- Include '@id's of candidate and ballot measure contest selections.
- Updates and bug fixes for write-ins in candidate contests.
- Multi-candidate slates:
  - Combine names of candidates into list.
    Names are stored in lists even when there is only one candidate.
  - Combine parties of candidate IFF they are the same.
    Parties are always stored in lists, and empty if there is no party.

Only updates the JSON file not the custom data file.

See: https://github.com/trustthevote-project/nist-1500-100-103-examples/test_cases/september_test_case.json

- Candidate parties are now a list.
- 'write_in' -> 'is_write_in'

These are NOT intended to follow any standard, merely to aid in debugging.
There's an ongoing conversation about the format of human readable IDs.

Use the September test case instead of the June test case, with order updated to match what's expected. Only handles Spacetown.
Human-readable IDs for ease of testing (not part of any agreed upon scheme).

Partly to share fields but primarily for correctness: contests have to be a
single list independent of sub-type because that's how EDFs work. Splitting
the types into multiple lists would lose the order.

…t styles.

…model.

Do some minor code cleanup.

An empty dictionary adds a special case to 'PartyData' construction.

- Unify extraction functions:
  - Use direct field accesses for properties, index for references.
  - Yield data dictionaries instead of returning lists.
- Add in election data.
- Entry point is now 'extract_ballot_data', which returns a list.

The caller only receives an 'ElectionData' back.

- Public API is now 'extract'.
- Prefix all internal methods as private.
  - Prefix utility functions but don't make them methods.
- Make the element index an instance variable.
- Re-wrap text to fit within line length.

Note: In a class method order usually puts public functions at the top.
Not doing that here to avoid making the change diff useless.

EDF to ballot data model extractor code is reviewed and approved.

Extraction code validates and generates ballots, passes GitHub Actions testing, but pytest still fails locally. Good enough for a merge.

* Update version to 0.1.3 upon milestone completion

* Update CHANGELOG.md

Co-authored-by: Ion Y <[email protected]>

* Update CHANGELOG.md

Co-authored-by: Ion Y <[email protected]>

* Update CHANGELOG.md

Co-authored-by: Ion Y <[email protected]>

* Use short dash in header

* Remove unneeded files

Co-authored-by: Ion Y <[email protected]>