Use iszero(?) instead of ?==zero(...) for warning when adding edge, for Symbolics Num compatibility

[kockahonza]

A partial fix/workaround for #49 with no downsides as far as I can tell. It is just a replacement of comparing to zero via == to using iszero. Though this isn't a comprehensive change to make all things work with Num types, it at least allows their creation (and they seem to mostly work in my usage).

[gdalle]

This seems fine to me, did you search for other uses of the pattern == zero( in the package while you're at it?

[kockahonza]

I've done a really quick grep for "zero" and didn't see it used anywhere else. Thanks for the promptness!

[gdalle]

Test failures are unrelated and should be fixed by #51

[simonschoelly]

Would it make sense to add Symbolics to the tests and write a small unit test so that this will not be accidentally removed in future releases?

[simonschoelly]

I am very unfamiliar with Symbolics otherwise I would have tested it myself but is this also a case we have to consider?

SimpleWeightedGraphs.jl/src/overrides.jl

Line 71 in 8128430

is_dangling = findall(S .== 0)

[gdalle]

Honestly I don't think it's worth the addition, Symbolics is a huge and rather brittle dependency. But we can try to come up with a simpler, self-contained test.

[gdalle]

@kockahonza sorry for the delay! Can you merge the latest main to this branch and add a self-contained test for your changes?