Add BoolVector2D/3D.

[nical]

This is something I have been missing for a while, not enough to go ahead and make the PR until now, but I think it is a nice addition.

This adds boolean vectors similar to what you can see in glsl and let you rewrite things like:

if some_vec3.x > another_vec3.x ||
  some_vec3.y > another_vec3.y ||
  some_vec3.z > another_vec3.z) {
  // do the thing!
}

into:

if some_vec3.greater_than(&another_vec3).any() {
  // do the thing!
}

We don't get to use the > operator because rust wants it to return a boolean, but I think it already makes things quite a bit nicer as is.
This PR doesn't implement all of the various candies we can think of for vector boolean ops, nor does it implement BoolVector4D but they can be added as followups if we need/want them.

---

This change is 

[nical]

r? anyone

[kvark]

First question, why are those separate types instead of just extra methods for TypedVector2D<bool, U>?

[nical]

I don't think the unit makes sense for these types, and the boolean vectors have a completely separate set of methods.

[kvark]

Second question is: what is the use case for actually needing BoolVector2D/3D as opposed to going straight to any()?

[kvark]

Needs some tests. Otherwise LGTM

[kvark]

that's just a subset of all possible swizzles. Is there a particular reason to only include those? Are those needed in practice?

[nical]

These are the swizzles we currently have in TypedVector2D. I wouldn't mind if we add the other ones but there are so many of them I'd rather just wait until someone asks for them.

I have used them a bit outside of webrender, but the one I use most at the moment is to go from 3d to 2d because foo = myvec.xy() is nicer and more explicit than foo = myvec.to_2d() in my opinion.
Having them doesn't add any real complexity so I don't think it's worth arguing over really.

[kvark]

yeah, myvec.xy() is nicer, but it's not clear why we shouldn't then be able to do myvec.yx()

[nical]

Totally. There is really no particular reason as far as I am concern to not add yx() and friends other than being busy with other things. If you write the PR I'll r+ it.

[kvark]

@nical alright, so how do you feel about adding some tests?

[nical]

Done.

[kvark]

Thanks!
what's up with the first commit?
otherwise r=me

[nical]

I'm tempted to say a github bug. It's the tip of the master branch, not sure why it shows up here since it is in master already. I tried to remove it and rebase on top of master but it still shpws up here. Anyway it rebases/merges cleanly so we can land now.

[KiChjang]

Hmm... the first commit hash differs from that of the commit found in the master branch... perhaps during rebase, it asked for you to recommit a commit that was in the master branch?

[nical]

I did:

git rebase -i @^^^
# removed the undesired commit
git pull --rebase upstream master

It didn't ask me anything in particular.
Git isn't my strong suit, I'll try to re-build the branch with cherry-pick instead of rebase to see if it works better.

[nical]

Looks like it worked.

[kvark]

@bors-servo r+

[bors-servo]

📌 Commit da7865a has been approved by kvark

[bors-servo]

⌛ Testing commit da7865a with merge a54f33a...

[bors-servo]

☀️ Test successful - status-travis
Approved by: kvark
Pushing a54f33a to master...