Fixes #1667 prefetching fails if clicking quickly back and forth

[ehrencrona]

Fixes #1667

When prefetching a page that is the current page, no props got stored in prefetching

I'm not entirely sure how to best write a test for this. Any pointers? Managed to add a test. It's a bit brittle (in the sense that if the prefetching logic changes it might pass even if the bug were to be reintroduced, not in the sense of sporadically failing) but I don't see any better way to handle it.

This is the kind of thing that unit tests might be better at, but the routing/preloading logic has so much global state that's pretty much impossible.

Before submitting the PR, please make sure you do the following

• It's really useful if your PR references an issue where it is discussed ahead of time. In many cases, features are absent for a reason. For large changes, please create an RFC: https://github.com/sveltejs/rfcs
• This message body should clearly illustrate what problems it solves.
• Ideally, include a test that fails without this PR but passes with it.

Tests

• Run the tests with npm test and lint the project with npm run lint

[benmccann]

I'm surprised lint isn't catching this, but it should probably be on the previous line: } else {

[benmccann]

Do we need to port this fix to Kit as well? https://github.com/sveltejs/kit/blob/master/packages/kit/src/runtime/internal/renderer/index.js

[ehrencrona]

Do we need to port this fix to Kit as well? https://github.com/sveltejs/kit/blob/master/packages/kit/src/runtime/internal/renderer/index.js

Good point. Probably, though the code has been rewritten so it's not obvious whether the issue still occurs. I'll try to find time to test that.

[ehrencrona]

Do we need to port this fix to Kit as well? https://github.com/sveltejs/kit/blob/master/packages/kit/src/runtime/internal/renderer/index.js

Good point. Probably, though the code has been rewritten so it's not obvious whether the issue still occurs. I'll try to find time to test that.

Update: I ported the test to Kit and it passes without any code changes, so I assume the fix does not need porting. Though I haven't investigated why in detail.