Fix ValueError from invalid CSS selectors in EmbedAPI (#12573)

This PR adds defensive handling for invalid CSS selectors in the
EmbedAPI to prevent crashes.

## Changes
- Added try-except blocks around  calls to catch  exceptions
- Invalid selectors now log a warning and return None instead of
crashing
- Applies to both the  query parameter and fragment-based selectors

## Issue
Fixes: https://read-the-docs.sentry.io/issues/6923723755/

Previously, when malformed CSS selectors like `@@R06GA` were passed,
selectolax would raise a ValueError that wasn't caught, causing the
entire request to fail.

Author: ericholscher

readthedocs/embed/v3/views.py

@@ -147,7 +147,11 @@ def _get_content_by_fragment(
 
     def _find_main_node(self, html, selector):
         if selector:
-            return html.css_first(selector)
+            try:
+                return html.css_first(selector)
+            except ValueError:
+                log.warning("Invalid CSS selector provided.", selector=selector)
+                return None
 
         main_node = html.css_first("[role=main]")
         if main_node:
@@ -183,7 +187,11 @@ def _parse_based_on_doctool(
             # characters as the `id=` HTML attribute
             # https://www.w3.org/TR/CSS21/syndata.html#value-def-identifier
             selector = f'[id="{fragment}"]'
-            node = HTMLParser(page_content).css_first(selector)
+            try:
+                node = HTMLParser(page_content).css_first(selector)
+            except ValueError:
+                log.warning("Invalid CSS selector from fragment.", fragment=fragment)
+                node = None
         else:
             html = HTMLParser(page_content)
             node = self._find_main_node(html, selector)