Generalize andM and similar functions

[gromakovsky]

Currently they only work on lists, but can be generalized I suppose.

[chshersh]

Okay, seems like those functions indeed can be generalized. This seems to work:

λ: let andM = foldr (\elem rest -> elem >>= \b -> if b then rest else pure False) (pure True) 
λ: :t andM
andM :: (Foldable t, Monad m) => t (m Bool) -> m Bool
λ: andM [Just True, Just False]
Just False
λ: andM [Just True, Just False, Nothing]
Just False
λ: andM [Just True, Nothing, Just False]
Nothing
λ: andM [putStrLn "1" >> pure True, putStrLn "2" >> pure False, putStrLn "3" >> undefined]
1
2
False

But don't know how it behaves with space leaks. Using for instead of whenJust introduces spaces leaks. That's why we have whenJust specialization. So I'm more cautious after that bug with such general definitions. @neongreen Could you check this definition on space-leakiness?

Also I don't know whether it's a good idea:

λ: andM (Just [True, False])
[ True , False ]
λ: andM (Just [False])
[ False ]
λ: andM (Just [False])

I expect some bugs after converting [Maybe Bool] into Maybe [Bool].

[neongreen]

I expect some bugs after converting [Maybe Bool] into Maybe [Bool]

This is not an issue if we use NontrivialContainer instead of Foldable.

Using for instead of whenJust introduces spaces leaks.

Instead of your definition I would just use the original list definition together with toList, to be safe.

[gromakovsky]

The only drawback is that version with toList is probably less efficient in some cases.

[chshersh]

We have #21 for that ;)
But I don't expect noticeable performance degrade.