Update no-var-keyword to dissalow var in export statements (fixes #1256)

[IllusionMH]

Fixes #1256

Test cases with extra whitespaces and comment added to ensure that new way to determine start of var keyword works correctly.

Also this is breaking changes and required update to mark one of previously existing test as failure (line 35).
One possible solution is to add argument "allow-in-exports" which will provide backward compatibility.

[adidahiya]

Nice, the code & tests lgtm @IllusionMH!

We've had this situation in the past (fixing a major false negative) and I think the approach has been to simply introduce the breaking change... but now that we are more mature maybe we should reconsider that. Thoughts @jkillian?

[jkillian]

If the rule is new and has just recently been added, I'm normally okay with giving it a release or two to fix false negatives since the amount of users is low. This seems like a rule that's widely used though, so I'd be worried about the build breaks this would cause.

On the other hand, I'm not against a major release, 4.0 soon if necessary.

[adidahiya]

@jkillian sure, I'm game for a 4.0 release soon; let's make sure to hit most of the other breaking changes for that.

In the meantime, let's try to get the other open PRs into a v3.15 release.