fix(jqLite): use .children to fetch child elements but fallback to .childNodes for SVGs in IE

[jbedard]

e35abc9 broke jqLiteDealoc on SVG elements in IE because .children is not supported on SVGs in IE (https://developer.mozilla.org/en-US/docs/Web/API/ParentNode.children#Browser_compatibility).

This change adds an SVG test and fixes it by doing .children || .childNodes. I also added .children in a few more places in addition to where e35abc9 originally did. Although if the .childNodes fallback is always required for IE, is it still worth using .children?

[mary-poppins]

Thanks for the PR! Please check the items below to help us merge this faster. See the contributing docs for more information.

• Uses the issue template (#8075)

If you need to make changes to your pull request, you can update the commit with git commit --amend.
Then, update the pull request with git push -f.

Thanks again for your help!

[jbedard]

@IgorMinar & @rodyhaddad since e35abc9 introduced the issue.

[jbedard]

Another option is to use .getElementsByTagName('*') instead of .children/.childNodes, that way no recursion is necessary. For single nodes .getElementsByTagName('*') is slightly slower, but for all other cases when the element has children it seems faster: http://jsperf.com/gebtn-vs-children

[rodyhaddad]

Thanks for making us aware of that!

Looking at jQuery's code, they use .getElementsByTagName when doing .remove. And as your jsperf shows, it's more performant.
So I think we're gonna go in that path for jqLite.

As for single nodes being slower, we can just add a el.childNodes.length check, which makes it a lot faster (see http://jsperf.com/gebtn-vs-children/2)

I'll make a PR for that, to hopefully get it in before next release

[rodyhaddad]

See #8136