FIX: Fitting free data use in clashom

[hulpke]

If the group has been obtained as subgroup from a fitting free/solvable radical computation,
the data will be inherited and might not guarantee that the factor group
really is fitting free. Test/resolve this.

This fixes #3139

Also recode assertion to avoid #3140

[codecov]

Codecov Report

Merging #3154 into master will increase coverage by 8.31%.
The diff coverage is 52.63%.

@@            Coverage Diff             @@
##           master    #3154      +/-   ##
==========================================
+ Coverage   75.36%   83.67%   +8.31%     
==========================================
  Files         682      688       +6     
  Lines      327430   336662    +9232     
==========================================
+ Hits       246757   281694   +34937     
+ Misses      80673    54968   -25705

Impacted Files	Coverage Δ
lib/grp.gd	100% <ø> (ø)	⬆️
lib/fitfree.gi	57.67% <100%> (+0.67%)	⬆️
lib/matrix.gi	86.63% <25%> (+10.03%)	⬆️
lib/clashom.gi	78.5% <50%> (+9.21%)	⬆️
src/vector.c	92.73% <0%> (-7.27%)	⬇️
src/hookintrprtr.c	67.34% <0%> (-4.88%)	⬇️
src/iostream.c	62.35% <0%> (-4.31%)	⬇️
src/vecffe.c	71.09% <0%> (-3.42%)	⬇️
src/lists.c	76.55% <0%> (-3.36%)	⬇️
src/c_type1.c	86.47% <0%> (-2.37%)	⬇️
... and 289 more

[coveralls]

Coverage decreased (-0.9%) to 82.832% when pulling 038098e on hulpke:fixes into 48b871b on gap-system:master.

[fingolfin]

Backported to stable-4.10 in commits 4f6aa5c fe8ebae 3f85d62

[olexandr-konovalov]

@hulpke for the release notes, could you please give a proper description of the fix? Thanks.

[hulpke]

@alex-konovalov I think the secription is as good as I can make it. For the release notes
Resolved an error in computing conjugacy classes of a subgroup
should be fine, but it also just cvould go under a general headline of numerous bugs were fixed, as the error was not dangerous.

[olexandr-konovalov]

I am particularly concerned about "fitting free computation" - seems this part could have been made clearer.

[hulpke]

@alex-konovalov
Fitting free/solvable radical is a class of algorithms that build the potentially problematic data structure. Are you concerned about the release notes (for which I suggested a line, and which do not need this detail), or the code base (for which I think the description in the PR is plausible?

[olexandr-konovalov]

About the style of the text used for release notes.

[olexandr-konovalov]

Let me explain in more details why I am assuming that for a side reader this may be cryptic.

First of all, it's now clear which is the behaviour before and which is after. Is what you're describing a bug, or a behaviour after a bug? Next,

If the group has been obtained as subgroup from a fitting free/solvable radical computation,

what is actually computed? A subgroup or a solvable radical? I would struggle to parse this

the data will be inherited and might not guarantee that the factor group

"will be" (after the fix) or "was" (before it) or actually "is" since this did not change at all?

really is fitting free. Test/resolve this.

and Fitting is a surname to be spelled as Fitting.

Would you be happy with this, or it's incorrect:

• #3154 If the group has been obtained as subgroup from a Fitting free/solvable radical computation, the data is inherited and might not guarantee that the factor group really is Fitting free. Added a check and an assertion to catch this situation.

[olexandr-konovalov]

I've put that version in #3503 - if you have any corrections, please leave them as comments there.