Add a second symmetric group presentation due to Moore

[MTWhyte]

This PR adds a second symmetric group presentation due to Moore, and therefore sets up the symmetric_group function to receive a third argument called index which can be used to choose between presentations of the same author.

In a separate commit, default value tests for the function cyclic_inverse_monoid are added.

[MTWhyte]

@james-d-mitchell a particular question I have is: do you think the way I've documented index is clear enough? I've tried to get across that an exception will be thrown if index is out of bounds, without writing a new\throws LibsemigroupsException specific to every single choice of author.

[MTWhyte]

I also considered adding the numbers of generators and relations for the symmetric group presentations while I was doing this. Since there are quite a lot of presentations (more than for cyclic_inverse_monoid and order_preserving_monoid), I wonder whether instead of having the list of bullet points with citations in brackets, we should have a table with fields author, index, reference, number of generators and number of relations?

[james-d-mitchell]

@james-d-mitchell a particular question I have is: do you think the way I've documented index is clear enough? I've tried to get across that an exception will be thrown if index is out of bounds, without writing a new\throws LibsemigroupsException specific to every single choice of author.

I think what you've got is clear, maybe you could combine the first and third throws and just say that it throws if the author index combination is invalid?

[james-d-mitchell]

I wonder whether instead of having the list of bullet points with citations in brackets, we should have a table with fields author, index, reference, number of generators and number of relations?

That'd be great, good idea.

[james-d-mitchell]

Also @MTWhyte would you mind rebasing this onto main, I made some changes to the CI. Thanks

[MTWhyte]

I've got the table working properly now. Including NOLINT as an html comment seems to have worked. A problem worth noting for future reference is, it appears that if the code for the table contains a line longer than about 160 characters, the table doesn't build properly.

[james-d-mitchell]

Thanks @MTWhyte