Increase Shogun coverage in benchmarks

[karlnapf]

This task is to check MLPack's benchmarking framework, and add more Shogun methods to it. We eventually want all of Shogun's algorithms in there to be covered.
This is really quite simple, and requires few scripts to be added, examples

Some information can be found here:
https://github.com/shogun-toolbox/shogun/wiki/GSoC_2017_project_fundamental_usual_suspects

Another nice thing to have would be a sorted list of where shogun does best/worst against other frameworks.

Steps:

1. Find out which algorithms are covered by the framework (both by shogun and in general)
2. Go to http://www.mlpack.org/benchmark.html and check the latest reports
3. Find a reported algorithm that is implemented in Shogun, but not yet covered by the benchmarks (LARS is such an example)
4. Add it
5. Goto 1

[pilar12]

I am interested in working on this. Which methods need to be worked on?

[karlnapf]

I have written a little list of steps to identify which method to work on above

[p16i]

I'm taking a look at this issue. I've created a spreadsheet for the current coverage at https://goo.gl/Qv9vxU.

I'll start go down the list, adding methods that are implemented in Shogun but aren't included in the benchmark.

Updated: An automatic way to get the results is using the cURL mentioned in #4097.

[karlnapf]

This is really useful info. Thanks for that! Curious to see what this leads to performance wise

[zoq]

Agreed, this is really nice.