`Error, collector is not reentrant` after previously interrupting code.

When working with PcGroups, I sometimes interrupt calculations because they take too long, e.g. ``` ^CError, user interrupt Stack trace: *[1] v[gn] := v[gn] + ew[sp]; @ /opt/gap/4.16.0-beta1/lib/rwspcsng.gi:189 [2] CollectWordOrFail( col, l, word ) @ /opt/gap/4.16.0-beta1/lib/rwspcclt.gi:491 [3] ReducedForm( rws, left * right ) @ /opt/gap/4.16.0-beta1/lib/rws.gi:353 [4] ReducedProduct( col, gk, ReducedProduct( col, gj, gi ) ) @ /opt/gap/4.16.0-beta1/lib/rwspcclt.gi:76 [5] FinitePolycyclicCollector_IsConfluent( col, false ) @ /opt/gap/4.16.0-beta1/lib/rwspcclt.gi:189 ... at *stdin*:8073 you can 'return;' ``` What *very* rarely happens, is that when I use `quit;` and then construct a new PcGroup and try to do calculations with it, I get hit with this: ``` Error, collector is not reentrant Stack trace: *[1] Error( "collector is not reentrant" ); @ /opt/gap/4.16.0-beta1/lib/rwspcsng.gi:98 [2] CollectWordOrFail( col, v, w ) @ /opt/gap/4.16.0-beta1/lib/rwspcclt.gi:425 [3] CollectWord( sc, av, y ); @ /opt/gap/4.16.0-beta1/lib/rwspcsng.gi:275 [4] SingleCollector_Solution( sc, gn[i], id ) @ /opt/gap/4.16.0-beta1/lib/rwspcsng.gi:793 [5] SingleCollector_MakeInverses( sc ); @ /opt/gap/4.16.0-beta1/lib/rwspcsng.gi:812 ... at *stdin*:8115 you can 'quit;' to quit to outer loop, or you can 'return;' to continue ``` Just running `quit;` again and restarting whatever calculation I was doing, tends to fix the problem. This has been happening for a long time (at least quite a few GAP versions), but it's very hard to recreate since it seems to rely on a user interrupt happening at a very specific moment. I suspect this is due to the interrupt happening in the middle of `SingleCollector_CollectWord`: https://github.com/gap-system/gap/blob/4a52d83616815bdd7dae6d9c8911a014ce49465b/lib/rwspcsng.gi#L74-L99 and the variable `SingleCollector_CollectWordRunning` never gets set to `false` due to the interruption. But because it gets set to `false` again before the `not reentrant` error is throwing, it works again from then onwards. This seems to be a way to enforce that no two instances of this function can run at the same time, but I don't immediately see *why* that should be prevented.