Rename bevy_reflect Array to Sequence

[stefanmitic]

Renamed bevy_reflect Array to Sequence and all associated types, traits and functions.
Fixed accompanying documentation, tests and examples.

Objective

Fixes #7059

Solution

Renamed Array to Sequence and made sure it is propagated where needed.

---

Changelog

Changed

bevy_reflect::Array to bevy_reflect::Sequence

[jakobhellermann]

We have List for "linear homogenous collection of elements, length can be changed" and Array for "linear homogenous collection of elements, length cannot be changed", right?

What makes Sequence a better name for the latter? Sequence sounds like something that both array and list would be, "iterable with fixed order, possibly no arbitrary indexing".

[soqb]

What makes Sequence a better name for the latter? Sequence sounds like something that both array and list would be, "iterable with fixed order, possibly no arbitrary indexing".

Array implies "length cannot be changed", but its length can be changed. Lists implement Array so a function taking dyn Array cannot be sure its length will never change.

i agree that Sequence is an awkward choice for the rename and agree with your definition. i'm advising against the renaming until we come up with a better alternative.

[jakobhellermann]

Array implies "length cannot be changed", but its length can be changed. Lists implement Array so a function taking dyn Array cannot be sure its length will never change

Good point. I guess this depends on how we interpret it. A list can be locally treated as an Array, which just restricts it to not using some methods. If you get a &mut dyn Array, you can be sure that while you have it nobody will change its length. Also with a &dyn Array, but you can have a &dyn List and a &dyn Array at the same time. But dyn Array is not a proof of the fact that the underlying array will have the same length at all points in time.
It's like how &T doesn't mean that the underlying storage will never get changed, just that it can be locally treated as read-only (until the lifetime runs out).

[MrGVSV]

I just left a comment on the original issue. I think another potential fix for the whole "fixed length" problem could be to just separate the two traits (and keeping the name Array).

[stefanmitic]

I think I can try and bang the code for the new approach out in the next few days, depending on free time.
Should we close this PR then?

[MrGVSV]

Should we close this PR then?

Maybe leave it open until the new PR is up so we can compare? Or at least until we get more feedback on the new issue?

[alice-i-cecile]

Closing in favor of the approach in #7121, which I think we agree is clearer.