[text-to-speech] Provide support for setting X-Watson-Learning-Opt-Out

[swgraham]

For many compliance related implementations, we need to use the X-Watson-Learning-Opt-Out Header on requests. Instantiating the text_to_speech with options should provide a documented and clear way to do it, for example:

watson.text_to_speech({
   username: blah,
   password: blah,
   voice: blah,
   opt-out: true,
});

Though I'm not sure on the tag we should use. You can work around this (it seems) by overriding the headers completely this way:

watson.text_to_speech({
   username: blah,
   password: blah,
   voice: blah,
   headers: {
        'content-type': 'application/json',
        'X-Watson-Learning-Opt-Out' : true }
   }
});

But you have to pass in the content-type or it will get overridden.

[nfriedly]

I just added in support for including X-Watson-Learning-Opt-Out as an option for the synthesize() method - does that sound like a good solution to you?

It should be listed in the jsdoc shortly.