saddle-points: add canonical data

[stkent]

Closes #582.

Notes:

• Per Intermediate functions: To test or not to test? discussions#41, the row/column identification tests found in several tracks represent testing intermediate implementation details and should be removed.
• Larger matrices have been removed from the tests; they did not exercise anything that was not already covered (except possibly for differing matrix dimensions?)
• Readme example is included as the first exercise; the majority of tracks already test this and it's a nice start point.
• Empty example is included; only a few tracks currently test this, but it's a nice boundary case.
• Multiple saddle points test uses the more-common matrix as found in e.g. xcsharp. xpython and xscala will need to update their multiple saddle points tests.
• Example with saddle point in the lower right matrix entry has been added to check for off-by-one errors. This is new for all tracks.
• Example with no saddle points has been expanded from 2x2 matrix to 3x3 matrix to give implementers more of a hint as to how one might go about constructing such a matrix. This is new for all tracks.
• No check for invalid matrix size is included in the canonical suite; can be included if desired, in which case I'll pull the xpython input for that test.

[ErikSchierboom]

Shouldn't this be null instead of an empty array? The README indicates it does:

if the input is valid but there is no result for the input, the value at "expected" should be null.

[petertseng]

In my interpretation, the empty result is a result, therefore it is not no result. [] makes more sense than null here.

[ErikSchierboom]

Okay, understood

[stkent]

Nice spot, my bad. On reflection, it would seem that this is a case where it makes sense to update the description; I'll push a commit with that change this weekend :o)

[stkent]

@ErikSchierboom forgive me, but I don't see where the line you quoted exists. The description for this exercise reads as follows:

A matrix may have zero or more saddle points.

Your code should be able to provide the (possibly empty) list of all the
saddle points for any given matrix.

Can you please point me to the source of the line you quoted? Thanks!

[ErikSchierboom]

@stkent It's in this repository's README. However, I think the empty list might after all be better in this case.

[stkent]

Ah, now I see, thanks! And yes, I think we now have three votes for an empty results list rather than an explicit null result :).

[ErikSchierboom]

Same as above

Not relevant anymore

[Insti]

Thanks for your work on this @stkent