feat: Implement Phase 2 Feed Formats - ATOM, JSON Feed, RSS fix (Phases 2.0-2.3)

This commit implements the first three phases of v1.1.2 Phase 2 Feed Formats,
adding ATOM 1.0 and JSON Feed 1.1 support alongside the existing RSS feed.

CRITICAL BUG FIX:
- Fixed RSS streaming feed ordering (was showing oldest-first instead of newest-first)
- Streaming RSS removed incorrect reversed() call at line 198
- Feedgen RSS kept correct reversed() to compensate for library behavior

NEW FEATURES:
- ATOM 1.0 feed generation (RFC 4287 compliant)
  - Proper XML namespacing and RFC 3339 dates
  - Streaming and non-streaming methods
  - 11 comprehensive tests

- JSON Feed 1.1 generation (JSON Feed spec compliant)
  - RFC 3339 dates and UTF-8 JSON output
  - Custom _starpunk extension with permalink_path and word_count
  - 13 comprehensive tests

REFACTORING:
- Restructured feed code into starpunk/feeds/ module
  - feeds/rss.py - RSS 2.0 (moved from feed.py)
  - feeds/atom.py - ATOM 1.0 (new)
  - feeds/json_feed.py - JSON Feed 1.1 (new)
- Backward compatible feed.py shim for existing imports
- Business metrics integrated into all feed generators

TESTING:
- Created shared test helper tests/helpers/feed_ordering.py
- Helper validates newest-first ordering across all formats
- 48 total feed tests, all passing
  - RSS: 24 tests
  - ATOM: 11 tests
  - JSON Feed: 13 tests

FILES CHANGED:
- Modified: starpunk/feed.py (now compatibility shim)
- New: starpunk/feeds/ module with rss.py, atom.py, json_feed.py
- New: tests/helpers/feed_ordering.py (shared test helper)
- New: tests/test_feeds_atom.py, tests/test_feeds_json.py
- Modified: CHANGELOG.md (Phase 2 entries)
- New: docs/reports/2025-11-26-v1.1.2-phase2-feed-formats-partial.md

NEXT STEPS:
Phase 2.4 (Content Negotiation) pending - will add /feed endpoint with
Accept header negotiation and explicit format endpoints.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

tests/helpers/__init__.py

@@ -0,0 +1 @@
+# Test helpers for StarPunk

tests/helpers/feed_ordering.py

@@ -0,0 +1,145 @@
+"""
+Shared test helper for verifying feed ordering across all formats
+
+This module provides utilities to verify that feed items are in the correct
+order (newest first) regardless of feed format (RSS, ATOM, JSON Feed).
+"""
+
+import xml.etree.ElementTree as ET
+from datetime import datetime
+import json
+from email.utils import parsedate_to_datetime
+
+
+def assert_feed_newest_first(feed_content, format_type='rss', expected_count=None):
+    """
+    Verify feed items are in newest-first order
+
+    Args:
+        feed_content: Feed content as string (XML for RSS/ATOM, JSON string for JSON Feed)
+        format_type: Feed format ('rss', 'atom', or 'json')
+        expected_count: Optional expected number of items (for validation)
+
+    Raises:
+        AssertionError: If items are not in newest-first order or count mismatch
+
+    Examples:
+        >>> feed_xml = generate_rss_feed(notes)
+        >>> assert_feed_newest_first(feed_xml, 'rss', expected_count=10)
+
+        >>> feed_json = generate_json_feed(notes)
+        >>> assert_feed_newest_first(feed_json, 'json')
+    """
+    if format_type == 'rss':
+        dates = _extract_rss_dates(feed_content)
+    elif format_type == 'atom':
+        dates = _extract_atom_dates(feed_content)
+    elif format_type == 'json':
+        dates = _extract_json_feed_dates(feed_content)
+    else:
+        raise ValueError(f"Unsupported format type: {format_type}")
+
+    # Verify expected count if provided
+    if expected_count is not None:
+        assert len(dates) == expected_count, \
+            f"Expected {expected_count} items but found {len(dates)}"
+
+    # Verify items are not empty
+    assert len(dates) > 0, "Feed contains no items"
+
+    # Verify dates are in descending order (newest first)
+    for i in range(len(dates) - 1):
+        current = dates[i]
+        next_item = dates[i + 1]
+
+        assert current >= next_item, \
+            f"Item {i} (date: {current}) should be newer than or equal to item {i+1} (date: {next_item}). " \
+            f"Feed items are not in newest-first order!"
+
+    return True
+
+
+def _extract_rss_dates(feed_xml):
+    """
+    Extract publication dates from RSS feed
+
+    Args:
+        feed_xml: RSS feed XML string
+
+    Returns:
+        List of datetime objects in feed order
+    """
+    root = ET.fromstring(feed_xml)
+
+    # Find all item elements
+    items = root.findall('.//item')
+
+    dates = []
+    for item in items:
+        pub_date_elem = item.find('pubDate')
+        if pub_date_elem is not None and pub_date_elem.text:
+            # Parse RFC-822 date format
+            dt = parsedate_to_datetime(pub_date_elem.text)
+            dates.append(dt)
+
+    return dates
+
+
+def _extract_atom_dates(feed_xml):
+    """
+    Extract published/updated dates from ATOM feed
+
+    Args:
+        feed_xml: ATOM feed XML string
+
+    Returns:
+        List of datetime objects in feed order
+    """
+    # Parse ATOM namespace
+    root = ET.fromstring(feed_xml)
+    ns = {'atom': 'http://www.w3.org/2005/Atom'}
+
+    # Find all entry elements
+    entries = root.findall('.//atom:entry', ns)
+
+    dates = []
+    for entry in entries:
+        # Try published first, fall back to updated
+        published = entry.find('atom:published', ns)
+        updated = entry.find('atom:updated', ns)
+
+        date_elem = published if published is not None else updated
+
+        if date_elem is not None and date_elem.text:
+            # Parse RFC 3339 (ISO 8601) date format
+            dt = datetime.fromisoformat(date_elem.text.replace('Z', '+00:00'))
+            dates.append(dt)
+
+    return dates
+
+
+def _extract_json_feed_dates(feed_json):
+    """
+    Extract publication dates from JSON Feed
+
+    Args:
+        feed_json: JSON Feed string
+
+    Returns:
+        List of datetime objects in feed order
+    """
+    feed_data = json.loads(feed_json)
+
+    items = feed_data.get('items', [])
+
+    dates = []
+    for item in items:
+        # JSON Feed uses date_published (RFC 3339)
+        date_str = item.get('date_published')
+
+        if date_str:
+            # Parse RFC 3339 (ISO 8601) date format
+            dt = datetime.fromisoformat(date_str.replace('Z', '+00:00'))
+            dates.append(dt)
+
+    return dates