feat: Add configurable timeout to FetchNode

[Xyerophyte]

Summary

This PR adds configurable timeout handling to FetchNode to prevent indefinite blocking on long-running HTTP requests and PDF parsing operations.

Problem

• Line 263 in fetch_node.py: requests.get(source) had no timeout parameter
• PDF parsing via PyPDFLoader.load() had no timeout mechanism
• Could lead to indefinite hangs on slow/unresponsive servers or large documents

Solution

1. Added Timeout Configuration

• New timeout attribute (default: 30 seconds)
• Configurable via node_config['timeout']
• Can be disabled by setting to None

2. Applied Timeout to HTTP Requests

• requests.get() now uses timeout parameter when configured
• Conditionally applied to preserve backward compatibility

3. Applied Timeout to PDF Parsing

• Wrapped PyPDFLoader.load() in ThreadPoolExecutor
• Enforces timeout and raises descriptive TimeoutError on timeout

4. Propagated Timeout to ChromiumLoader

• Automatically passes timeout to ChromiumLoader via loader_kwargs
• Respects explicit loader_kwargs['timeout'] if already set

Usage Examples

# Default 30s timeout
node = FetchNode(input="url", output=["doc"], node_config={})

# Custom timeout
node = FetchNode(input="url", output=["doc"], node_config={"timeout": 15})

# Disable timeout (legacy behavior)
node = FetchNode(input="url", output=["doc"], node_config={"timeout": None})

Testing

• Added comprehensive unit tests in tests/test_fetch_node_timeout.py
• Tests cover: default/custom/disabled timeout, HTTP requests, PDF parsing, ChromiumLoader propagation

Backward Compatibility

✅ Fully backward compatible:

• Existing code works without changes (gets 30s timeout)
• Legacy behavior available by setting timeout=None
• No breaking changes

Related Issue

Addresses the issue mentioned in line 263 of fetch_node.py where requests could block indefinitely.