[FIX] core: inconsistent pypdf metadata in 5.1

Even if 5.1 is not technically supported, debuntu might switch to it
any moment and we don't know if / when upstream will fix it, so it's
not worth the time bomb.

While technically it was always typed such, since py-pdf/pypdf#2820 it
looks like `_info` is a lot more likely to be `None` as
e.g. `clone_reader_document_root` now starts with unsetting
`_info_obj` and never re-sets it.

Except `add_metadata` was not updated to handle this case, likely
because mypy interprets `assert isinstance(self._info,
DictionaryObject)` as a type narrowing and trusts the developer, thus
does not report the type mismatch... and the assertion ends up blowing
in the user's face at runtime with a simple

    >>> r = pypdf.PdfReader(some_pdf_document)
    >>> w = pypdf.PdfWriter()
    >>> w.clone_reader_document_root(r)
    >>> w.add_metadata({"/foo": "bar"})
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "lib/python3.12/site-packages/pypdf/_writer.py", line 1622, in add_metadata
        assert isinstance(self._info, DictionaryObject)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    AssertionError

This is rather inconsiderate, so monkeypatch `add_metadata` to handle
the case where `_info` exists and is `None`.

Most of the credit goes to juwu for uncovering the issue.

opw-4372052
opw-4426881
Fixes odoo#185673

X-original-commit: 1fb35ad
Co-authored-by: Junqi <juwu@odoo.com>

odoo/tools/pdf/_pypdf.py

@@ -1,3 +1,5 @@
+from typing import Dict, Any
+
 import pypdf
 from pypdf import errors, filters, generic, PdfReader as _Reader, PdfWriter as _Writer
 from pypdf.generic import create_string_object
@@ -45,6 +47,11 @@ def getFormTextFields(self):
 
 
 class PdfWriter(_Writer):
+    def add_metadata(self, infos: Dict[str, Any]) -> None:
+        if hasattr(self, '_info') and self._info is None:
+            self._info = generic.DictionaryObject()
+        super().add_metadata(infos)
+
     def getPage(self, pageNumber):
         return self.pages[pageNumber]