Error about "Relation not homogeneous"

[fingolfin]

Mike Newman reported an issue in anupq 3.1.1 via email, when running the test program below (warning: running this may take multiple hour, and produces more than 6 GB of log data). To quote Mike:

Running the attached input file reports a (in fact many) Relation not homogeneous.
as an attached extract from the output file shows.
In the process of preparing this report I noticed that in the final structure table
some of the right-hand sides are not normal words.
There are some squares at the end and sometimes a final term .1^0 or ..0^0.

The corresponding input to Magma V2.21-2 on the same machine
does not give the problem.
I don't know what standalone version the Magma process is based on.

LoadPackage("anupq");
SetInfoLevel(InfoANUPQ, 6);
procId := PqStart( FreeGroup(6),1000000000 : Exponent := 4, Prime := 2 );
i := procId;
PqPcPresentation(procId : ClassBound := 5, OutputLevel := 1);
PqSetMaximalOccurrences( i, [3,3,3,3,3,3] );

for c in [6..8] do
  PqComputePCover(i);
  PqDoExponentChecks(i);
  PqEliminateRedundantGenerators(i);
od;

#class 9
PqComputePCover(i);
PqDoExponentChecks(i: Bounds := [1,8]);
PqEliminateRedundantGenerators(i);

#class 10
PqComputePCover(i);
PqDoExponentChecks(i: Bounds := [1,8]);
PqEliminateRedundantGenerators(i);

PqSetOutputLevel(3);

#class 11
PqSetupTablesForNextClass(i);
PqTails(i, 11);
PqDoConsistencyChecks(i, 11, 0);
PqTails(i, 10);
PqDoConsistencyChecks(i, 10, 0);
PqTails(i, 9);
PqDoConsistencyChecks(i, 9, 0);
PqTails(i, 8);
PqDoConsistencyChecks(i, 8, 0);
PqTails(i, 7);
PqDoConsistencyChecks(i, 7, 0);

PqDisplayStructure(i);

PqQuit(procId);

quit;

[fingolfin]

An excerpt from the log:

#I  Jacobi was 131 9 1
#I  The quotient relation is  .13929 .13938 .14310 .14723 .16897 .17495 .19858 .28331 .28344 .28372 .28381 .28460 .29024 .29041 .2
9822 .30143 .30144 .32338 .41735 .41744 .41768 .41777 .41819 .41820 .41967 .41982 .41998 .41999 .42012 .42542 .42621 .42668 .42669
 .42773 .43372 .45434 .45435 .54060 .54249 .54265 .54279 .54616 .55163 .55266 .55284 .55455 .55470 .56093
#I  Jacobi was 132 9 1
#I  Jacobi was 133 9 1
#I  Jacobi was 134 9 1
#I  Jacobi was 135 9 1
#I  Relation not homogeneous of class 11. Relation ignored.
#I  Jacobi was 136 9 1
#I  Relation not homogeneous of class 11. Relation ignored.
#I  Jacobi was 137 9 1
#I  Relation not homogeneous of class 11. Relation ignored.
#I  Jacobi was 138 9 1