This PR configures the npm test command in the client-ts\FunctionalTests to automatically launch a browser (headless chrome, by default) and run the functional tests, collecting the results and writing them out.

There are a few elements to this:

• We register a Jasmine Reporter in the Functional Tests that generates Test-Anywhere Protocol (TAP) output describing the test run as <li> elements in a hidden <ul> element (optionally made visible by the displayTap query string value)
• The run-tests.js script, uses the selenium-webdriver package. It:
  1. Launches Selenium WebDriver
  2. Launches the server
  3. Instructs WebDriver to open the page on the server that runs the tests in a browser (any browser supported by Selenium WebDriver)
  4. Scrapes the DOM, reading the hidden <li> values and write them to the console, until the tests complete
  5. Terminates the browser, server and WebDriver.
• Now, the run-tests.js script basically acts as a script that writes TAP output for our functional tests. We can pipe that through any TAP-compatible reporter (we use faucet here) to report output.

IMPORTANT: This is not yet integrated into our build, the first pass is just to make it runnable and provide the infrastructure.

Also, this mechanism should be able to automate BrowserStack/Sauce Labs/some other browser lab. We'll need to work out the mechanisms for deploying the server code somewhere, but we should be able to use this to move forward with that.

Examples:

Successful test run (click to embiggen):

Failed test run (click to embiggen):

TODO

• Rebase
• Fix tslint errors (thanks @BrennanConroy :(... but seriously it's good)
• Don't build the entire client in FunctionalTests.csproj, we can make sure it builds before we run the tests in the CI but doing it in FunctionalTests.csproj is bogging down VS builds.