Ready for review and testing . I will add later some comments about changes

Unfortunately, it doesn't seem to pass CI, but the issue #1344 is fixed with this PR

Codecov Report

Base: 94.62% // Head: 94.62% // Increases project coverage by +0.00% 🎉

Coverage data is based on head (53aad08) compared to base (e23b985).
Patch coverage: 100.00% of modified lines in pull request are covered.

@@           Coverage Diff           @@
##             main    #1345   +/-   ##
=======================================
  Coverage   94.62%   94.62%           
=======================================
  Files          30       30           
  Lines        5118     5137   +19     
  Branches     1052     1058    +6     
=======================================
+ Hits         4843     4861   +18     
  Misses        164      164           
- Partials      111      112    +1     

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

more details on mod:
a) _cmap.py : the str is coming from "/Encoding" which stores a NameObject : the conversion is already performed no need to force it.
b) _page.py : Obsolete called missed in _debug_for_extract()
c) _base.py:
unnumber : all #xx should be performed prior to conversion to str (using utf-8) to allow multi language text:

read_from_stream : if utf-8 (normally the only one required) or gbk (kept to prevent regression) we will use charmap to get some sequence of chars
renumber : added to recode in #xx sequence :

renumber will also be compatible with utf-8 chars

else test added (some test were not iaw pdf spec and have been corrected)
a test using the pdf from the issue has been added in test_merger

Thank you @pubpub-zz 🙏

I will make a release to PyPI in a few minutes.