Attributes: fix setting selected on an option in IE<=11

[timmywil]

• Same as the getter fix. All that was required was accessing the selectedIndex property.
• Also added support for optgroups

Fixes gh-2732

[mgol]

What about Edge? I'd love all new IE-specific stuff to be tested on Edge and if the bug is not present there to write IE <= 11 only.

Adding spaces around <= would be good as well, that's where the conventions seem to go currently.

[timmywil]

Edge is ok. Updated comment, but the syntax says there's no space after the <=, so it's

IE <=11 only.

[gibson042]

Can the get and set bodies be made even more similar? gzip will love you...

[dmethvin]

Going even further, can get and set use the same method?

[timmywil]

Going even further, can get and set use the same method?

I suppose they can, but notice that the getter doesn't access selectedIndex on the select unless there's an optgroup. That said, if we're okay with it, I'll just make the getter and setter the same function. I don't think it'll hurt anything to go back to accessing selectedIndex twice in the getter.

[dmethvin]

Yeah, I doubt an extra access will make that much difference. Since it's already guarded by a support test, only IE will pay that penalty anyway.

[timmywil]

Actually, they can't be the same function. The getter needs to return null and the setter needs to return undefined. I tried a factory to create similar functions, but that was +15 bytes. Even using the same function (even though it wouldn't work right) was +4 bytes. Just for kicks, I tried making the getter identical to the setter – with the extra return statement, and it was exactly the same. I think what was there is still the best option.