Add optional wordvec feature for RelationshipSourceCollection

[SOF3]

Objective

• Describe the objective or issue this PR addresses.
• If you're fixing a specific issue, say "Fixes #X".

Add a more memory-efficient relationship source collection data structure for relationships with up to one child most of the time.

Solution

• Describe the solution used to achieve the objective above.

Introduces a new optional dependency wordvec, under the feature wordvec, such that

WordVec<Entity, 1>

can be used to substitute

SmallVec<[Entity; 1]>

where the former only takes 16 bytes instead of 24 bytes with comparable performance to SmallVec, particularly 20%-40% when used as Vec<WordVec<Entity, 1>> to index other entities, but 40% slower during single-item pushing.

Accessing Vec<X<usize>> for X = Vec/ThinVec/SmallVec/WordVec (with N = 1) where vec.len() == 1000 && vec[i].len() == 1 for i in 0 to 1000 (effectively Query<&Children>.iter().flatten().for_each(|e| *u64_query.get_mut(e).unwrap() += 1)):

Pushing 1, 2, 3 in order to X<u16> with N = 3:

Testing

• Did you test these changes? If so, how?
• Are there any parts that need more testing?
• How can other people (reviewers) test your changes? Is there anything specific they need to know?
• If relevant, what platforms did you test these changes on, and are there any important ones you can't test?

The tests for SmallVec implementation in relationship_source_collection.rs are duplicated for WordVec as-is.

[SOF3]

Not very familiar with the bevy CI system, not sure if any changes are required to enable the wordvec feature in the CI by default.

[alice-i-cecile]

I'm generally a bit skeptical about pulling in more dependencies for stuff like this, but the implementation is fine. Very annoying that orphan rule makes us decide on this upstream.

[SOF3]

It is also possible to implement this in the wordvec crate as well, but my intuition is that wordvec is supposd to be lower-level than bevy.

[alice-i-cecile]

It is also possible to implement this in the wordvec crate as well, but my intuition is that wordvec is supposd to be lower-level than bevy.

Agreed, bevy is a better home for this than wordvec.