fix(model): accept relative URLs in re_sign_file_url_answer

Regression for #40788. The new multi-pattern URL detector required an
`https?://` prefix, which broke the existing tests that use relative
`/files/...` URLs inside markdown link payloads.

- Make the host prefix optional so all three supported shapes (markdown,
  backticked, bare) match both relative and absolute paths.
- Replace the trailing `.*` after each URL parameter with a
  bounded `[^)\s]*?` so the bare-URL pattern doesn't run off the
  end of the answer string and gobble up adjacent URLs.

api/models/model.py

@@ -1669,7 +1669,13 @@ class Message(Base):
         # The original implementation only matched the markdown form, so
         # bare and backticked tool file URLs kept the long-lived
         # INTERNAL_FILES_URL host and 5xx-ed at serve time. Refs #40788.
-        url_core = r"https?:\/\/.+?\/files\/(tools\/)?[\w-]+.*?timestamp=.*&nonce=.*&sign=.*"
+        # The host prefix is optional so relative `/files/...` URLs that
+        # the agent returns without a host are also covered. The
+        # `(?=[)\s`]|$)` at the end of the bare-URL pattern (and the
+        # closing backtick / paren on the wrapped forms) stops the
+        # greedy `.*?=.*?` after `&sign=` from running off the end of
+        # the answer.
+        url_core = r"(?:https?:\/\/.+?)?\/files\/(tools\/)?[\w-]+.*?timestamp=[^)\s]*?&nonce=[^)\s]*?&sign=[^)\s]*?"
         patterns = [
             r"\[!?.*?\]\((" + url_core + r")\)",  # [text](url)
             r"`(" + url_core + r")`",              # `url`
@@ -1683,9 +1689,6 @@ class Message(Base):
         if not urls:
             return self.answer
 
-        if not urls:
-            return self.answer
-
         re_sign_file_url_answer = self.answer
         for url in urls:
             if "files/tools" in url:
@@ -1728,6 +1731,7 @@ class Message(Base):
                 result = re.search(upload_file_id_pattern, url)
                 if not result:
                     continue
+
                 upload_file_id = result.group(1)
                 if not upload_file_id:
                     continue