feat: add "markdown" field type to extraction strategy schema (#1708)

[hafezparast]

Summary

• Adds "markdown" as a new field type in JsonElementExtractionStrategy's type pipeline
• Converts selected element's HTML to clean markdown (preserves bold, links, lists, headers) without returning raw HTML tags
• Works across all strategy subclasses: JsonCssExtractionStrategy, JsonLxmlExtractionStrategy, JsonXPathExtractionStrategy
• Composable in pipelines: ["markdown", "regex"] works naturally

Closes #1708

Changes

• crawl4ai/extraction_strategy.py: 8 lines — add elif step == "markdown" branch + CustomHTML2Text import
• tests/test_markdown_field_type_1708.py: 16 tests covering CSS/lxml/XPath strategies, pipeline chaining, edge cases, JSON serialization

Usage

from crawl4ai.extraction_strategy import JsonCssExtractionStrategy

schema = {
    "name": "articles",
    "baseSelector": "div.article",
    "fields": [
        {"name": "title", "selector": "h1", "type": "text"},
        {"name": "body", "selector": "div.content", "type": "markdown"},  # NEW
        {"name": "author", "selector": "span.author", "type": "text"},
    ],
}

strategy = JsonCssExtractionStrategy(schema)

Before (only options):

Type	Output
"text"	Best seller - Our most popular widget with advanced features.
"html"	<p><strong>Best seller</strong> - Our most popular widget with <em>advanced features</em>.</p>

After (new option):

Type	Output
"markdown"	**Best seller** - Our most popular widget with _advanced features_.

Also works in pipelines:

# Extract markdown, then regex a pattern from it
{"name": "highlight", "selector": "p.desc", "type": ["markdown", "regex"], "pattern": r"\*\*(.+?)\*\*"}

Test plan

• 16 unit tests passing (pytest tests/test_markdown_field_type_1708.py -v)
• Existing pipeline tests pass with no regressions (pytest tests/test_pr_1290_1668.py -v)
• Verified across JsonCss, JsonLxml, and JsonXPath strategies
• Verified JSON serialization of results works cleanly

🤖 Generated with Claude Code