Add ES6 Symbol support

[meeber]

Changes:

• Fix bug when testing Symbol equality with should syntax ([ES6] should.equal fails for Symbols #669)
• Add Symbol tests for assert, expect, and should

Notes:

• No symbol tests were added to the key-checking block because of a separate bug. Will open a new issue on it soon. The bug occurs because both the expected and actual array of keys were being sorted in the assertKeys function using Array.prototype.sort but this throws a TypeError when one of the keys is a Symbol in a Map or a Set.
• No Symbol tests were added to the NaN block because JavaScript's isNaN() throws a TypeError when passed a Symbol. I don't know if Chai is supposed to mimic this behavior with its NaN test or instead return true?
• No Symbol tests that intentionally cause an error (and then test the string) were added. This is because the currently published version of Mocha doesn't correctly display AssertionError messages containing Symbols. However this issue appears to be resolved in Mocha's master branch; at least it displays "{}" to represent a Symbol.
• I was unable to use the more succinct typeof this === 'symbol' approach mentioned in [ES6] should.equal fails for Symbols #669 because this is wrapped in an object and thus returns 'object' as its typeof.

[meeber]

On second thought, this Map/Set fix might be problematic.

If a user already has a keys-related test written that uses Map or Set (without any Symbol keys), and if it relies on the parameters in the AssertionError message to be sorted via Array.prototype.sort, then I believe their test would stop passing.

It's gonna be a little awkward to respect the existing behavior of sorting the expect/actual in the AssertionError message for Map and Set key checks but not sort them if either contains keys that are Symbols.

[keithamus]

Hey @meeber thanks for the PR!

In the nicest way possible; try not to do too much in one PR. If you dial this back to just the fix discussed in #669 and then we can work on how to address the keys issue separately - without holding up the fix to Symbol equality. Sound good?

[meeber]

Agree 100%. Was actually doing that when you posted.

The changes have been made, only #669 fix and tests are in now.

[keithamus]

In that case, fantastic work @meeber 😄

[keithamus]

LGTM. @lucasfcosta?

[meeber]

Gotta run to work now so I'll open issues for the other stuff later. Thanks!

[lucasfcosta]

LGTM.
Nice work and thanks for pointing out other issues.