trim doesn't trim enough

... as in this example:
```

i1 : R=QQ[x]

o1 = R

o1 : PolynomialRing

i2 : M=coker matrix {{1,x,0},{0,0,1}}

o2 = cokernel | 1 x 0 |
| 0 0 1 |

2
o2 : R-module, quotient of R

i4 : M=coker matrix {{x,x^2,0},{0,0,1}}

o4 = cokernel | x x2 0 |
| 0 0 1 |

2
o4 : R-module, quotient of R

i6 : N = trim M

o6 = subquotient (| 1 |, | 0 x |)
| 0 | | 1 0 |

2
o6 : R-module, subquotient of R

i9 : P = subquotient ( map(R^2, R^1, {{1},{0}}), matrix {{x},{0}} )

o9 = subquotient (| 1 |, | x |)
| 0 | | 0 |

2
o9 : R-module, subquotient of R

i11 : g = inducedMap ( N,P )

o11 = | 1 |

o11 : Matrix

i12 : isIsomorphism g

o12 = true
```

HT: @eisenbud

Start by reproducing the Macaulay2 transcript, especially the trim M result and the induced isomorphism to P. The issue names no source file or test, so locate the implementation and coverage for trim before determining what behavior should change; done means the reported example has a justified trim result.

Written by the indexing model from the issue text.