I often get failures when running all tests in parallel because this test takes more than 1 second to run.
Example:
> failed - expected to be close to <2022-05-24 10:59:18.3370> (within 1), got <2022-05-24 10:58:18.0000>
To allow the test to take a few more seconds, I think this is a reasonable compromise.
This adds a third case `disabled` that allows tests to explicitly disable StoreKit 2.
Some types, like `TrialOrIntroPriceEligibilityChecker` and `TransactionsManager` always used SK2 because the implementation was better. But this meant that tests couldn't explicitly disable it (other than calling the internal methods directly).
This allows #1464 to be a lot simpler but being able to use the public `Purchases` methods, but forcing the implementation to use SK1.
Note that this is purely an implementation detail. `StoreKit2Setting` is `internal`, and the public API still only exposes the same boolean.
This basically solves [CF-497] too, since those tests will now force `TrialOrIntroPriceEligibilityChecker` to run the SK1 implementation.
[CF-497]: https://revenuecats.atlassian.net/browse/CF-497?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ
* made a few attributes that aren't present in SKTest receipts optional, added test with IAP
* added more tests for every field in the receipt
* fixed a few places where `asDict` wasn't correctly considering nullability in the IAP receipt fields
* added basic ugly test
* fixed local receipt parsing for sktest, broke regular receipt parsing temporarily
* updated length calculation so that it works for both storekit test and regular receipts
* updated test to allow for missing OriginalAppVersion in receipt
* fixed linting in imports
* fixed checks for non-storekitTest receipts
* added a longer comment to explain the length == 0 case
* cleaned up the still-ugly tests for the receipt parser
NachoSoto