normal form circuits not equivalent to input circuit

[the-entire-country-of-ireland]

The normal form circuits do not appear to be equivalent to the input circuit. To test this, I randomly generated stabilizer circuits C, and then simulated the output of running NF(C) C^\dagger using a stabilizer simulator library. This should give the all-zero state, but it instead gives a mix of different outcomes.

I have attached an explicit small circuit for which this mismatch
circ.txt
occurs.