std: Make std.PriorityQueue and std.PriorityDequeue unmanaged containers

[saurabh-mish]

#23431 had been dormant for a while.

This PR revives #21433 and resolves #21432.

[rohlem]

Unrelated to these changes, but it seems theres'a typo in the doc comment:
PriorityDeque says Order.gt if the third argument should be min-popped second,
while Queue says Order.gt if the third argument should be min-popped first.
From my understanding PriorityQueue is consistent here, the error for Deque was made in 22e1d92, but not in corresponding 82e8930.

(The wording is a bit clunky imo, I'd personally just replace the last part with "otherwise", or drop it since it's logically implied - but up to you or maintainers to decide.)

[saurabh-mish]

Summary of changes

Priority Dequeue

std.PriorityDequeue has method signatures similar to std.Deque

• Moved methods removeMinOrNull into removeMin and removeMaxOrNull into removeMax; updated tests.

• Renamed methods add, removeMin and removeMax to push, popMin and popMax repectively.

• Added new method isEmpty which returns true if the priority dequeue is empty and false if not. It is being used in addUnchecked, peekMin, removeMin, and removeMax.

Moved Priority Dequeue into Priority Deque

Priority Queue

std.PriorityQueue has method signatures similar to std.PriorityDequeue

• Moved method removeOrNull into remove and updated corresponding tests.

• Renamed methods add to push and remove to pop.

• Added new method isEmpty method which returns true if the priority queue is empty and false if not. It is being used in peek and pop methods.

• Renamed the cap field to capacity (as present in ArrayList) and renamed the capacity method to getCapacity

TODO (this part will be edited based on progress, feedback, and observation)

• Add / improve doc comments
  • PriorityQueue
  • PriorityDeque
• Consistent fields: PriorityQueue has items, cap, and context, whereas PriorityDeque has items, len, context
  • Was struggling with this to be honest. If this is indeed an issue, I think it is beyond the scope of this PR.
• .empty value
  • PriorityQueue
  • PriorityDeque

[linux-user36]

Since there's already deque.zig, should this maybe be renamed to PriorityDeque for consistency?

[saurabh-mish]

Since there's already deque.zig, should this maybe be renamed to PriorityDeque for consistency?

I think so too, pushing PriorityDeque related changes shortly

[imomaliev]

IMHO we should not move simple checks like self.len > 0 to a function. It would have been needed if the empty state was complex or required some additional checks. Because it is just a length check, by adding this helper function we are actually making code slower, but without substantial readability improvements.

http://ithare.com/wp-content/uploads/part101_infographics_v08.png

[rohlem]

Small functions can (and are expected to) be trivially inlined by the compiler's optimization pass.

[saurabh-mish]

Adding an isEmpty method was intentional because its commonly used in general purpose programming. After adding it, I thought of actually using it in existing methods.

[mrjbq7]

I don't think an isEmpty method is worth adding, they don't exist for other containers and len checks are simple.

[mrjbq7]

Also weird implementation that doesn't just return self.len == 0;

[saurabh-mish]

Fixed isEmpty methods of both containers

[saurabh-mish]

I don't think an isEmpty method is worth adding, they don't exist for other containers and len checks are simple.

I feel that methods which provide information about a field are easier to read than directly accessing it.

For example, a method count already exists in both these containers, which is easier to read than items.len.
Similarly, the newly added isEmpty method is easier to read than items.len == 0.

[saurabh-mish]

This PR is ready for review (updated the summary of changes).