Fix draft listing for current Substack API responses

[giacomo-folli]

Api.get_drafts() previously returned the entire response object, while callers expected a list of draft records. This caused drafts list and other draft-management operations to handle the response incorrectly.

This PR extracts the draft records from the posts field while preserving compatibility with responses that already return a list.

Details

The drafts API response format includes pagination metadata alongside the draft records. Returning the complete dictionary from Api.get_drafts() breaks consumers that expect to iterate over draft objects.

For example, the CLI expects to process each draft with calls such as:

draft.get("id")
draft.get("title")

When the response envelope is returned directly, those callers receive the dictionary itself instead of the list under posts.

Implementation

Api.get_drafts() now:

1. Handles the API response through the existing response helper.
2. Detects responses containing a list-valued posts field.
3. Returns that list to preserve the existing method contract.
4. Falls back to returning the original response for compatibility with older or alternative response formats.

Test Coverage

Added a regression test that verifies:

• A response containing posts, hasMore, and nextCursor is handled correctly.
• Api.get_drafts() returns only the draft records.
• Existing draft-list behavior remains compatible with callers expecting a list.

Documentation

Added an Unreleased changelog entry describing the fix.

The formatting hooks also normalized existing formatting in the widget-preservation implementation and test files. These changes are formatting-only and do not alter their behavior.

[ma2za]

Reviewed the response normalization and all callers. Returning the posts list preserves the CLI and internal list contract while retaining legacy list responses. Offline tests, the merge result against current main, package builds, and CI across Python 3.10–3.14 pass.