update inductive_canonical_cycle test

[lcnr]

my concern was already known 😅

chalk/book/src/recursive/inductive_cycles.md

Lines 48 to 66 in 1c2c0ca

	## What happens if we treat inductive cycles as errors?
	So, what do we do when we hit an inductive cycle? Given that we told you that an
	inductive proof cannot contain cycles, you might imagine that we can just treat
	such a cycle as an error. But this won't give us the correct result.
	Consider our previous example. If we just treat that cycle as an error, then we
	will conclude that the impl for `Vec<T>` doesn't apply to `?X: A`, and we'll
	proceed to try the impl for `u32`. This will let us reason that `?X: A` is
	provable if `?X = u32`. This is, in fact, correct: `?X = u32` *is* a possible
	answer. The problem is, it's not the only one!
	In fact, `Implemented(?X: A)` has an **infinite** number of answers. It is true
	for `?X = u32`. It is true for `?X = Vec<u32>`. It is also true for
	`Vec<Vec<u32>>` and `Vec<Vec<Vec<u32>>>` and so on.
	Given this, the correct result for our query is actually "ambiguous" -- in
	particular, there is no unique substitution that we can give that would make the
	query provable.

r? @jackh726

[jackh726]

@bors r+

[bors]

📌 Commit 0ffedf2 has been approved by jackh726

It is now in the queue for this repository.

[bors]

⌛ Testing commit 0ffedf2 with merge 808257c...

[bors]

☀️ Test successful - checks-actions
Approved by: jackh726
Pushing 808257c to master...