Restore 'browser: false' to Brackets core's JSLint setup

[peterflynn]

Fix Brackets project settings so browser: true is not set (which is how we had it before 834c4e9 in Sprint 37). Fix a few JSLint errors that snuck in because of this, plus some additional errors in ExtensionLoader that seem to have come from the Themes work in 0.42

[dangoor]

@peterflynn But, with the exception of our node code, most of Brackets runs in a browser environment where window and other common globals can be expected to exist. Why do we not want to assume a browser environment?

[peterflynn]

@dangoor We have it here in the old coding standards: https://github.com/adobe/brackets/wiki/Brackets-Coding-Conventions#globals. I think the main issue is that JSLint's browser option includes too much stuff -- it lets you access top-level vars named event, name, parent, etc., all of which are likely to mask bugs if ever encountered in our code. Even document is an issue since we often use local vars named document to reference Document objects.

I'd be ok with adding window plus setTimeout & friends to the default JSLint globals, though -- if there's an easy way to do it.

[peterflynn]

Looks like we could add "predef": ["window", "setTimeout", "clearTimeout"] to the "jslint.options" pref for our project.

@dangoor How far do you think we should take that? We could also add define & brackets (and maybe even $) to the list, which could let us entirely eliminate the /*global*/ declaration from most of our files. Worth doing?

[marcelgerber]

@peterflynn Mustache too, for the sake of completeness ;)
(there may even be some others)

[dangoor]

@peterflynn Oh right, document is a particularly good example of a variable that appears in the browser but that we'd rather pretend doesn't exist because we use that name, too.

I like the idea of setting browser: false and adding predefs as you suggest.

[nethip]

@peterflynn LGTM. Is "predef": ["window", "setTimeout", "clearTimeout"] the only change that is pending for this PR to go to master?

[MiguelCastillo]

Why are we removing window from this one?

[MiguelCastillo]

@peterflynn I really like this idea... It will remove most the jslint definitions in source files, which I personally think isn't the best place to add those settings.

Why haven't we merged this in yet?

[ficristo]

In #11693 I am converting Brackets core code to ESLint.
Want to land this patch before it?

[ficristo]

Fixed in #12661