Batch constraints to within maxTupleSize in deriveArgDict

[jonathanlking]

This PR fixes #47 by batching constraints to never exceed maxTupleSize.

[Ericson2314]

Can you add a test with a large sum type?

[Ericson2314]

I think it might be better to do something like

Suggested change

	batchConstraints :: [Type] -> Type
	batchConstraints cs
	| length cs == 0 = TupleT 0
	| otherwise = AppT (AppT (TupleT 2) initialBatch) (batchConstraints rest)
	where
	(initial, rest) = splitAt (maxTupleSize - 1) cs
	initialBatch = foldl AppT (TupleT (length initial)) initial
	batchConstraints :: [Type] -> Type
	batchConstraints cs
	| length cs <= maxTupleSize = foldl AppT (TupleT (length cs)) cs
	| otherwise = batchConstraints $ batchConstraints <$> chunkBy maxTupleSize cs

maybe there is an issue, but that seems about right. I am out of time to think about this ATM.

[jonathanlking]

That seemed to work and I'm very happy to make the change. I couldn't find a chunkBy, or any function :: Int -> [a] -> [[a]], in base or an existing dependency, so I just wrote an implementation in this library rather than add a new library dependency.

[jonathanlking]

This appears to have been fixed by #35.
I have rebased this branch, just keeping the test. There might be value in merging the test to prevent a future regression, but I wouldn't be too disappointed otherwise.

[ali-abrar]

Looks like the test doesn't pass