feat(agent): scrape linked pages to ground answers in real content

Adds `_fetch_page_content()`, which downloads a URL with a
browser-like User-Agent and uses BeautifulSoup with the lxml
parser to strip scripts, navigation, and other non-content
markup, returning up to 3000 characters of readable text.

The response validator now also compares generated replies
against this scraped text, flagging named entities and claims
that do not appear in the retrieved page content.

Dependencies: adds beautifulsoup4==4.12.3 and pins lxml==5.3.0
for parser compatibility.

venv/lib/python3.11/site-packages/lxml/proxy.pxi

@@ -71,9 +71,7 @@ cdef xmlDoc* _plainFakeRootDoc(xmlDoc* c_base_doc, xmlNode* c_node,
             return c_base_doc
 
     c_doc  = _copyDoc(c_base_doc, 0)                   # non recursive!
-    c_new_root = tree.xmlDocCopyNode(c_node, c_doc, 2) if c_doc is not NULL else NULL # non recursive!
-    if c_new_root is NULL:
-        raise MemoryError()
+    c_new_root = tree.xmlDocCopyNode(c_node, c_doc, 2) # non recursive!
     tree.xmlDocSetRootElement(c_doc, c_new_root)
     _copyParentNamespaces(c_node, c_new_root)
 
@@ -576,10 +574,7 @@ cdef void fixThreadDictNamesForDtd(tree.xmlDtd* c_dtd,
                 _fixThreadDictPtr(&c_element.content.prefix, c_src_dict, c_dict)
             c_attribute = c_element.attributes
             while c_attribute:
-                if tree.LIBXML_VERSION < 21500:
-                    # libxml2 2.15 no longer stores default values in the dict.
-                    # See https://gitlab.gnome.org/GNOME/libxml2/-/commit/24628f25
-                    _fixThreadDictPtr(<const_xmlChar**>&c_attribute.defaultValue, c_src_dict, c_dict)
+                _fixThreadDictPtr(&c_attribute.defaultValue, c_src_dict, c_dict)
                 _fixThreadDictPtr(&c_attribute.name, c_src_dict, c_dict)
                 _fixThreadDictPtr(&c_attribute.prefix, c_src_dict, c_dict)
                 _fixThreadDictPtr(&c_attribute.elem, c_src_dict, c_dict)