Bug in `new_nac_homs!`

[kris-brown]

The logic of new_nac_homs! is to find, after a deletion $f: X \leftarrowtail X'$, new morphisms which may have arisen due to a NAC $n: L \rightarrow N$ being deactivated.

This is done via finding morphisms $h: N\rightarrow X$ that meet some property, then pulling back $n \cdot h$ along $f$ to obtain the match $L\rightarrow X'$. However, something is awry in defining "some property", since the code presently allows $h$ to map parts of $L$ into deleted parts of $X$, which make the pulled back morphism $L\rightarrow X'$ undefined. So that logic needs to be looked into + improved.

[kris-brown]

I think one issue here might be the fact that, for SPO rewriting, the effect of the deletion can extend beyond the context of where the rule was explicitly applied. It would be good to check whether this bug can affect DPO rules or just SPO rules - perhaps there is a less efficient strategy to take when the rule applied is SPO.