Improve InverseMatMod with integer modulus

[fingolfin]

Before:

gap> m2:=RandomUnimodularMat(100);;
gap> for i in [1..100] do x:=InverseMatMod(m2,2); od; time;
277
gap> for i in [1..100] do x:=InverseMatMod(m2,251); od; time;
1564
gap> for i in [1..100] do x:=InverseMatMod(m2,65537); od; time;
Error, reached the pre-set memory limit

After:

gap> m2:=RandomUnimodularMat(100);;
gap> for i in [1..100] do x:=InverseMatMod(m2,2); od; time;
273
gap> for i in [1..100] do x:=InverseMatMod(m2,251); od; time;
1249
gap> for i in [1..100] do x:=InverseMatMod(m2,65537); od; time;
1226

[codecov]

Codecov Report

Merging #2426 into master will decrease coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #2426      +/-   ##
==========================================
- Coverage   73.88%   73.88%   -0.01%     
==========================================
  Files         484      484              
  Lines      245462   245444      -18     
==========================================
- Hits       181368   181350      -18     
  Misses      64094    64094

Impacted Files	Coverage Δ
lib/zlattice.gi	77.27% <100%> (-0.32%)	⬇️
src/objset.c	84.59% <0%> (-0.23%)	⬇️
hpcgap/lib/hpc/stdtasks.g	64.45% <0%> (+0.25%)	⬆️

[hulpke]

Yes, avoiding to pre-calculate the multiplication table should perform better, in particular if the prime is larger.

[fingolfin]

It also avoids exhaustin the complete RAM ;-)