Add sum-of-multiples.json

[petertseng]

This is a synthesis of the existing tests in current tracks.

A short attempt to discern the history of the various tests:

Almost all tracks have the [3, 5] cases in common, as well as the
[7, 13, 17] and [43, 47] case. Indeed, these cases were present since
the first known sighting of sum-of-multiples at
exercism/exercism@607be68.

The [3, 5] to 100 case was added after discussion in
https://github.com/exercism/todo/issues/20 and
exercism/exercism#2486. Most tracks had
adopted them at the time of this PR.

The [4, 6] and [5, 6, 8] cases arose from discussion in
exercism/exercism#2341 - some tracks, but
not all, had adopted these tests at the time of this PR.

The [5, 25] case was a Haskell-specific case added in
exercism/haskell#50.

The [] and [1] cases were Go-specific cases that have been present ever
since the exercise was added to the Go track in
exercism/go#53.

Cases found in various tracks that were not included:

The [1] case of Go was originally a [1, 1] case but this is a conscious
decision not to test duplicate factors.

There were [0] and [0, 1] cases added to the Python track in
exercism/python#149 but this is a conscious
decision not to test 0 as a factor.

This is the second subtask of #198.

Closes https://github.com/exercism/todo/issues/154

[petertseng]

Discussion is welcomed on whether to include tests for:

• Having duplicates in the list of factors
• Having 0 in the list of factors

[petertseng]

Verified no mistakes in this file:

require 'json'

cases = JSON.parse(IO.read('sum-of-multiples.json'))

cases['cases'].each { |c|
  expected = c['expected']
  factors = c['factors']
  limit = c['limit']
  observed = (1...limit).select { |n|
    factors.any? { |f| n % f == 0 }
  }.reduce(0, :+)

  raise "Nope on #{c}, got #{observed}" if observed != expected
}

puts 'pass'

[kytrinyx]

I think that it's the right move to not test duplicate factors. Also, 0 is a really weird case, with the identify for multiplication being 1 and I'd rather not complicate this exercise with it.

I think that this PR is good as is, and would love a second (and third) opinion.

[petertseng]

Thanks for review. Just FYI I'm changing the commit message (the json file contents haven't changed) so as to reference https://github.com/exercism/todo/issues/20, as I missed it in my history exploration.

We'll want to change the README (first subtask of #198) when tracks start adopting these, but that can wait a bit (and is an easy change to make in any case). Will probably emblazon the tracks; about time I gave that a try.

[kytrinyx]

Excellent! Let me know if you run into any weirdness with blazon.