[Groff] [Heirloom] Paragraph adjust "badness" calculation

[carsten . kunze]

Hello,

address@hidden (https://github.com/reffort) did notice a major issue in one of 
the key features of Heirloom roff which is paragraph-at-once formatting.  
Gunnar very likely did implement this algorithm based on the 1981 Knuth-Plass 
paper (KP-81 in the following text) and had maybe not been aware that TeX had 
implemended a slightly different algorithm later after gaining more experience 
in that field (TeX82 in the following text).  But even KP-81 is implemented 
incorrectly to such a degree, that paragraph formatting in current Heirloom is 
noticeable far from being optimal.  The full story of reffort's analysis and 
change proposal can be read at 
https://github.com/n-t-roff/heirloom-doctools/issues/22

There are two issues in Heirloom with that.  One is the scaling of the 
deviation from the ideal interword space size (which allows too tight lines).  
A second issue is the use of that scaled value in the badness and demerits 
calculation.  Here Gunnar maybe did misinterpret KP-81, the implementation is 
obviously wrong.

reffort did not only provide a solution for these issues, he also provided a 
test document and the possibility to process this document with the algorithm 
described in KP-81, TeX82, two new proposals and current Heirloom (see 
https://github.com/n-t-roff/heirloom-doctools/issues/22).

Heirloom users are encouraged to test these algorithms (also with their own 
documents) so a final decision regarding the changes could be made.

--Carsten