feat(sources): add RSS/Atom feed adapter with content extraction

[foundatron]

Closes #5

Changes

1. tentacle/sources/rss.py — Add HTML-to-text parser and optional content extraction

• Add _HTMLToTextParser(html.parser.HTMLParser) class (stdlib) to strip HTML tags and extract plain text. Skip content inside <script>, <style>, <nav>, <header>, <footer>. Collapse whitespace.
• Add _fetch_content(url: str, timeout: int, max_bytes: int) -> str | None module-level function:
  • Fetches via urllib.request.urlopen with the given timeout.
  • Reads up to max_bytes via resp.read(max_bytes).
  • Detects charset from Content-Type header, falls back to UTF-8 with errors="replace".
  • Parses HTML to text via _HTMLToTextParser.
  • Catches all exceptions, logs warning, returns None on failure.
• Add __init__(self, extract_content: bool = False, content_timeout: int = 30, content_max_bytes: int = 1_048_576) to RSSAdapter.
• In _parse_rss and _parse_atom, after building each Article: if self._extract_content is enabled and the article has a URL, call _fetch_content and set article.full_text.

2. tentacle/config.py — Add RSSSourceConfig with extract_content field

• Create RSSSourceConfig(SourceConfig) dataclass with extract_content: bool = False.
• Change Config.rss field type from SourceConfig to RSSSourceConfig (default factory RSSSourceConfig).
• Update _apply_toml() to read extract_content from the [sources.rss] TOML dict and apply it to the config, following the same pattern used for HN's min_points, arXiv's days_back, etc.
• Update DEFAULT_CONFIG_TEMPLATE to add # extract_content = false comment under [sources.rss].

3. tentacle/cli.py (~line 85–86) — Pass extract_content to RSSAdapter

• Change RSSAdapter() to RSSAdapter(extract_content=config.rss.extract_content).

4. config.example.toml — Add extract_content comment under [sources.rss]

• Add # extract_content = false line.

Review Findings

• Errors: 0
• Warnings: 3
• Nits: 4
• Assessment: NEEDS CHANGES

The most actionable fix is #1 (log the exception). Items #2 and #3 are worth addressing before this ships to production — #2 for operational sanity and #3 to avoid leaking script content on malformed pages.