add package NumericalSemigroups

[eisenbud]

New package created by Frank Schreyer and David Eisenbud with basic commands for analyzing semigroups using
the Kunz cones
and studying when they are Weierstrass semigroups.

[eisenbud]

Also updated "PencilsOfQuadrics"

[eisenbud]

many tests in Pencil... used random numbers and thus failed randomly. I've inserted setRandomSeed commands to fix this.

[d-torrance]

The random seed is always set to 0 by default for tests and examples, so I don't think the most recent commit is necessary.

The failing build from earlier was unrelated to this PR (#1579).

[eisenbud]

Thanks, Doug -- This is my first experience(s) doing a pull request by myself. I had assumed that the failing sumsOfSquares call indicated in the "Details" was something indirectly called from my package. Should I have just ignored the failure?
-David

[d-torrance]

Often build failures are related to the PR (like the Homepage one earlier), so I wouldn't say you should ignore them in general.

If you're not sure, then you can search through the existing issues to see if the failure is a known problem like this one, or just ask!

[mahrud]

Usually if only a single build fails it could be one of these unrelated but persistent failures that show up randomly, but if two fail probably one of the tests failed and four fail it's probably a failing example.

Perhaps at some point we can have GitHub annotations #3533 to easily tell us what failed without having to dig through the build log.

[d-torrance]

@eisenbud - Did you want to keep the setRandomSeed commit? If not, you can run the following:

git reset --hard HEAD~
git push --force <remote> development

HEAD~ refers to the commit immediately preceding the current one. Replace <remote> with whatever name you've given your fork.

[mikestillman]

@d-torrance @eisenbud The one failing test (I think) is the SumOfSquares failure, which should not be related to this PR.
Our Complexes PR (#3526) will make changes to PencilOfQuadrics too. This PR should be done before #3526, that way, I can deal with the resulting conflicts.

[d-torrance]

The one failing test (I think) is the SumOfSquares failure, which should not be related to this PR.

That's correct -- see #1579. @mahrud has a fix in #3519

I'm not sure if this should be merged now or not. The final commit (3e672a6) may not be necessary.

[mahrud]

David, do you mind if I force push a changes to your branch to remove the last commit and fix the failing build?

[mahrud]

This line is going to cause errors (and also the "restart" two lines above will prevent this test from running)

[d-torrance]

The build failure is unrelated (#3239). Merging!