fix(plex): plain text survives a decimal reference too long to parse

int() refuses a decimal run past 4300 digits, so one such reference in a summary raised out of the sync and the text migration. Any number past seven significant digits is beyond U+10FFFF, so it becomes U+FFFD like any out-of-range reference; leading zeros are dropped first. The rails test's comment now says what it guards: both rails agree with the rollup.

backend/app/plex/plaintext.py

@@ -27,7 +27,14 @@ _REF = re.compile(r"&(?:#[0-9]+|#[xX][0-9a-fA-F]+|[A-Za-z][A-Za-z0-9]*);")
 
 def _decode(m: re.Match[str]) -> str:
     ref = m.group()
-    return html.unescape(ref) if ref[1] == "#" or ref[1:] in html5 else ref
+    if ref[1] != "#":
+        return html.unescape(ref) if ref[1:] in html5 else ref
+    if ref[2] in "xX":
+        return html.unescape(ref)
+    # `int()` refuses a decimal run past 4300 digits, and any number past seven significant digits is
+    # beyond U+10FFFF, which the parser turns into U+FFFD anyway.
+    digits = ref[2:-1].lstrip("0") or "0"
+    return "�" if len(digits) > 7 else html.unescape(f"&#{digits};")
 
 
 def plain_text(value: str | None) -> str | None: