[jimklimov]

Follows up from #1945

@xdelaruelle : Seems the long-words policy for TextWrapper() does it, thanks for the hint. Now the python2/3 generated files are almost identical, and differ a lot from those generated without this change (no longer broken at slashes and dashes in the path names).

 diff -bu ChangeLog*n
--- ChangeLog2n 2023-06-02 12:11:58.987738723 +0200
+++ ChangeLog3n 2023-06-02 12:12:10.797738492 +0200
@@ -22992,12 +22992,11 @@

 2014-02-14  Arnaud Quette <arnaud.quette@free.fr>

-       * docs/acknowledgements.txt: Update NUT team membership for
-         Frédéric Bohe  Frederic Bohe, NUT senior developer and Eaton
-         contractor from 2009 to 2013, is now a retired member. Thanks for
-         all the hard work on the Windows port, nut-scanner, Unix packaging,
-         support, ... Also update the developers membership page, from
-         Alioth to GitHub
+       * docs/acknowledgements.txt: Update NUT team membership for Frédéric
+         Bohe  Frederic Bohe, NUT senior developer and Eaton contractor from
+         2009 to 2013, is now a retired member. Thanks for all the hard work
+         on the Windows port, nut-scanner, Unix packaging, support, ... Also
+         update the developers membership page, from Alioth to GitHub

 2013-02-24  Charles Lepple <clepple+nut@gmail.com>

@@ -23319,8 +23318,8 @@
          environment on ships (and we don't need shutdown there -- in
          critical situations the system has to operate as long as possible,
          untill the battery is empty)  Also, all of the previous issues
-         listed below are now fixed in this al175 version:  - ‘return’
-         with a value, in function returning void (2x) - anonymous variadic
+         listed below are now fixed in this al175 version:  - ‘return’ with
+         a value, in function returning void (2x) - anonymous variadic
          macros were introduced in C99 - C++ style comments are not allowed
          in ISO C90 - ISO C forbids braced-groups within expressions (5x) -
          ISO C90 forbids specifying subobject to initialize (16x) - ISO C99

Nearby context of both changes seems to be non-ASCII unicode, so maybe that's why they wrapped differently.