Update sinon to 1.17.2

[tylerball]

I mainly want this for https://github.com/sinonjs/sinon/issues/702 (in version 1.14.0), which fixes async tests when using sinon.test wrapping a mocha test, but there's lots of other useful stuff in here:

1.17.0 / 2015-09-22
==================

  * Fix #821 where Sinon.JS would leak a setImmdiate into global scope
  * Removed sinon-timers from the build. refs #811
  * Added flag that, when set to true, makes sinon.logError throw errors synchronously.
  * Fix #777: Support non-enumerable props when stubbing objects
  * Made the sinon.test() function pass on errors to the callback
  * Expand conversion from ArrayBuffer to binary string
  * Add support for ArrayBuffer, blob responseTypes

1.16.1 / 2015-08-20
===================
* Bump Lolex to stop throwing an error when faking Date but not setTimeout

1.16.0 / 2015-08-19
===================
* Capture the stack on each spy call
* fakeServer.create accepts configuration settings
* Update Lolex to 1.3.0
* Fire onreadystatechange with event argument
* Returns supersedes previous throws
* Bunch of bug fixes

1.15.0 / 2015-05-30
==================
* Fixed bug where assertions don't handle functions that have a property named proxy
* update license attribute
* Add test coverage report
* responseHeaders on abort are empty object
* Fix pre-existing style error
* Update documentation to cover testing built version
* Update CONTRIBUTING.md with section about "Making a pull request"
* Improve RELEASE.md to reduce effort when cutting a new release
* Deprecate mock
* Release.md
* Make `npm docs sinon` work.
* Run unit tests against packaged version in CI environment
* Remove unused Gruntfile
* Use Vanilla Buster.JS
* Use `files` in package.json
* Fix code style
* Don't stub getter properties
* Event listeners for `progress`, `load` and `readystatechange` in the `readyStateChange` function in  `FakeXMLHttpRequest` are dispatched in a different order in comparison to a browser. Reorder the events dispatched to reflect general browser behaviour.
* Update linting instructions in CONTRIBUTING.md
* Lint all files with new linter
* Update JSCS to 1.11.3 and make npm lint task verify all files
* Cleanup .restore()

== 1.14.1 / 2015-03-16
* Fallback for .restore() native code functions on Chrome & PhantomJS (なつき)
* Restore support for sinon in IE<9 (Harry Wolff)

== 1.14.0 / 2015-03-13
* Stub & spy getters & setters (Simon Zack)
* Fix #702 async sinon.test using mocha interface (Mohayonao)
* Add respondImmediately to fake servers (Jonathan Freeman)

== 1.13.0 / 2015-03-04
* fix @depends-require mismatches (fixes AMD issues) (Ben Hockey)
* Fix spy.calledWith(undefined) to return false if it was called without args
* yieldsRight (Alexander Schmidt)
* stubs retain function arity (Charlie Rudolph)
* (AMD) use explicit define in built version
* spy().reset() returns this (Ali Shakiba)
* Add lengthComputable and download progress (Tamas Szebeni)
* Don't setContent-type when sending FormData (AJ Ortega)
* sinon.assert with spyCall (Alex Urbano)
* fakeXHR requests in Node. (Jmeas)
* Enhancement: run builds on docker (till@php.net)
* Use FakeXDomainRequest when XHR does not support CORS. Fixes #584 (Eric Wendelin)
* More lenient check for ActiveXObject
* aligned sandbox.useFakeXMLHttpRequest API to documentation (Phred)
* Fix #643. Returns supersedes previous throws (Adam Hull)
* Safely overwrite properties in IE - no more IE files!
* Add check for setInterval/clearInterval (kdpecker)
* Add safety check for document.createElement (kdpecker)
* Fix #633. Use a try/catch when deleting a property in IE8. (Garrick Cheung)

== 1.12.1 / 2014-11-16
* Fixed lolex issue on node

== 1.12.0 / 2014-11-16
* Fake timers are now extracted as lolex: http://github.com/sinonjs/lolex
* Improved setImmediate fake
* Proper AMD solution

== 1.11.1 / 2014-10-27

* Expose match on returned sandbox (Duncan Beevers)
* Fix issue #586 (Antonio D'Ettole)
* Declare log_error dependency (Kurt Ruppel)

== 1.11.0 / 2014-10-26

* Proper AMD support
* Don't call sinon.expectation.pass if there aren't any expectations (Jeffrey Falgout)
* Throw error when reset-ing while calling fake
* Added xhr.response property (Gyandeep Singh)
* Fixed premature sandbox destruction (Andrew Gurinovich)
* Add sandbox reset method (vitalets)
* A bunch of bug fixes (git log)
* Various source organizational improvements (Morgan Roderick and others)

== 1.10.3 / 2014-07-11

* Fix loading in Web Workers (Victor Costan)
* Allow null as argument to clearTimeout and clearInterval (Lars Thorup)

== 1.10.2 / 2014-06-02

* Fix `returnValue` and `exception` regression on spy calls (Maximilian Antoni)

== 1.10.1 / 2014-05-30

* Improved mocha compatibility for async tests (Ming Liu)
* Make the fakeServer log function overloadable (Brian M Hunt)

== 1.10.0 / 2014-05-19

* Ensure that spy createCallProperties is set before function invocation (James Barwell)
* XDomainRequest support (Søren Enemærke, Jonathan Sokolowski)
* Correct AMD behavior (Tim Branyen)
* Allow explicit naming of spies and stubs (Glen Mailer)
* deepEqual test for unequal objects in calledWithExactly (Bryan Donovan)
* Fix clearTimeout() for Node.js (Xiao Ma)
* fix fakeServer.respond() in IE8 (John Bernardo)
* Fix #448 (AMD require.amd)
* Fix wrapMethod error handling (Nikita Litvin)

== 1.9.1 / 2014-04-03

* Fix an issue passing `NaN` to `calledWith` (Blake Israel)
* Explicate dependency on util package (Kris Kowal)
* Fake timers return an object with `ref` and `unref` properties on Node (Ben Fleis)

== 1.9.0 / 2014-03-05

* Add sinon.assert.match (Robin Pedersen)
* Added ProgressEvent and CustomEvent. Fixes bug with progress events on IE. (Geries Handal)
* prevent setRequestHeaders from being called twice (Phred)
* Fix onload call, 'this' should be equal to XHR object (Niklas Andreasson)
* Remove sandbox injected values on restore (Marcus Hüsgen)
* Coerce matcher.or/and arguments into matchers (Glen Mailer)
* Don't use buster.format any more
* Fix comparison for regexp deepEqual (Matt Kern)

[jejacks0n]

Looks good to me. Can you please confirm that the broken specs aren't related to changing sinon? They may be trying to load an incorrect version now -- if you think they're unrelated, we'll look at them internally and merge this as-is.

[jejacks0n]

Never mind, it would seem they are not related based on your commit.

[jejacks0n]

The 500 is probably related to https://github.com/modeset/teaspoon/issues/443