Implement StandardAssociate(Unit) for ZmodnZ, tweak documentation for IsEuclideanRing

[fingolfin]

Another PR coming from "spring cleaning" in my local branches. Adjusting the documentation for IsEuclideanRing is necessary because for e.g. polynomials, the euclidean degree can be -infinity.

Having StandardAssociate and StandardAssociateUnit for ZmodnZ simply is useful for writing various generic algorithms.

[codecov]

Codecov Report

Merging #1990 into master will decrease coverage by 0.02%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1990      +/-   ##
==========================================
- Coverage   66.01%   65.98%   -0.03%     
==========================================
  Files         898      898              
  Lines      273347   273634     +287     
  Branches    12766    12770       +4     
==========================================
+ Hits       180439   180554     +115     
- Misses      90078    90249     +171     
- Partials     2830     2831       +1

Impacted Files	Coverage Δ
lib/zmodnz.gi	95.57% <100%> (+0.15%)	⬆️
hpcgap/lib/zmodnz.gi	95.08% <100%> (+0.16%)	⬆️
lib/float.gi	29.89% <0%> (-1.77%)	⬇️
src/macfloat.c	75.25% <0%> (-0.57%)	⬇️
src/hpc/thread.c	45.81% <0%> (-0.4%)	⬇️
src/funcs.c	76.37% <0%> (-0.27%)	⬇️
src/gap.c	58.32% <0%> (-0.16%)	⬇️
src/stats.c	79.43% <0%> (ø)	⬆️
src/stringobj.c	79.14% <0%> (+0.11%)	⬆️
lib/ring.gi	51.93% <0%> (+0.22%)	⬆️
... and 4 more

[olexandr-konovalov]

Looks good, although I think it will be useful not only to check the output of each function, but also to check that the condition StandardAssociateUnit( r ) * r = StandardAssociate( r ) holds.

[fingolfin]

Excellent idea, and it revealed a bug in the code -- the unit it computed was the reciprocal of the correct one :-/. This is fixed, and additional tests have been added.