crypto-square: Remove test of intermediate functions

[behrtam]

So, I was just looking at the canonical test data for crypto-square and I think it was the first time I saw test data for an exercise that also tested the intermediate steps of the problem like normalization or concatenation of segments.

Shouldn't those intermediate tests be removed, as it was agreed on in: exercism/discussions#41

They originated from this PR: #250

[Insti]

Link to canonical-data.json

[petertseng]

Indeed, the Haskell track discussed in exercism/haskell#259 and decided not to test intermediate functions, diverging from x-common in exercism/haskell#391.

Not saying that all have to take the exact approach the Haskell track took or that any other track should agree with the exact same reasons, but if this data helps make the decision, we are glad to provide it.

[petertseng]

Since crypto-square's introduction into the Go track at exercism/go#68 the Go track has never tested the intermediate functions. So if x-common goes this route, xgo would become closer to x-common.

[nholden]

#922 was merged, so I think we can close out this issue.