Add a MonadBaseControl instance for PropertyT

[treeowl]

I believe it is possible to write a lawful MonadBaseControl instance
for CofreeT (see this proof), and therefore also for the TreeT, GenT, and
PropertyT types in this package. I can't say for sure whether it
"makes sense", but hey, maybe it does.

[jacobstanley]

Wow this is awesome, I have wanted this for a long time, I didn't think it was possible

[jacobstanley]

I can't say I really understand how this works, but given it won't break any existing code I say we give it a try. Not sure what would be the best way to check this works.

[treeowl]

Thanks! As I said, I don't really know if this instance is reasonable—the MonadBaseControl laws smell somewhat incomplete to me, so I don't know just what to look for in that regard. But here's hoping!

[jacobstanley]

I've just been trying this out and it does appear to work, at least for the case of using bracket_ from lifted-base. Time will tell I suppose, but I am about to recommend that people use lifted-base for bracketing from now on given ResourceT has gone the UnliftIO route which we'll never be able to implement.

This has saved us from being stuck on an old ResourceT forever! 🎉 🙏 thanks again

[ocharles]

Now that we have MonadBaseControl PropertyT is there any need to have TestT?

[jacobstanley]

That's a good question, it would be interesting to see what happens with lifted-async !

This was the original reason for TestT if I recall:
https://github.com/hedgehogqa/haskell-hedgehog/blob/master/hedgehog/src/Hedgehog/Internal/State.hs#L819-L837

It would be nice to see if the linearize + parallel testing functionality could be made to work with the lookupIx style stateful tests.