feat(runtime): cache queries with useSWR

[ghost]

Implements client side caching with useswr (https://swr.vercel.app). See https://jira.dhis2.org/browse/LIBS-166 for the related epic.

Notes:

• Some types need to be fixed, ignored some of them for the moment
• Currently not cancelling requests. useSWR doesn't have an explicit API for this yet: Cancel requests vercel/swr#129, though there are possibilities apparently
• We're not using the arguments passed by swr to the queryFn (engine.query in our case) at the moment. Though it should be possible to use the arguments that swr passes us.
• Refetch does not return a promise currently, so this is incompatible with how our refetch used to work. We'll need to see if we want to either drop this behaviour, or implement it with useSWR with the new refetch.
• onComplete and onError will fire for each request. Even when swr is serving cached data. Could be problematic. It might mess with app logic when swr returns (stale) data and an error callback fires.
• SWR also supplies onSuccess and onError hooks to pass a callback to. We could use those as well.
• The cache key for a group of queries could change if a single query of the group of queries changes, even though it should only affect the cache key of the changed query. This means that currently caching isn't as efficient with multiple queries for one hook. There doesn't seem to be anything native to swr that allows multiple queries easily like react-query's useQueries.
• Unlike react-query, useSWR uses referential equality checks to check whether the cache key has changed.

Tests:

• onComplete, onError callbacks
• variables (initial)
• multiple queries
• refetch (for lazy and non-lazy)

API docs:

• https://swr.vercel.app/docs/options

Follow-up (unrelated to swr):

• Remove waitForElement
• Address act warnings (everything in renderHook should already be wrapped in act)

[ghost]

Closed in favor of #824