Export namespaced browser global and support config

[danny0838]

Currently this module exports lots of global variables for the browser, this may be undesired and inflexible.

This change exports everything under a global namespaced object parseCss instead for the browser (Use "Css" to prevent conflict with a potential official API, which generally uses "CSS"). For other platforms like Node.js/CommonJs/AMD there is no difference.

Another commit adds a config object to allow the user to configure the behavior of this module, such as a custom error handle for the tokenizer and parser, which can resolve issue #41. An example usage look like:

var parseErrorHandler = parseCss.config.parseErrorHandler = function (index, token) {
  parseErrorHandler.errors.push(
    `Parser error at index ${index}, token ${token} on css: ${cssText}`;
  );
};

parseErrorHandler.errors = [];
var cssText = /* input css string */;
var tokens = parseCss.tokenize(cssText);
// do something with parseErrorHandler.errors

Though this is still sub-optimal since a manual cleanup and a memory of the input is required before each call. For a more elegant implementation I would recommend refactoring the tokenize() and parse*() methods to be classes, so that it can be like:

const {Tokenizer: CssTokenizer} = parseCss;

const tokenizerErrorHandler = function (tokenizer, input, index, token) {
  tokenizer.errors = tokenizer.errors || [];
  tokenizer.errors.push(
    `Parser error at index ${index}, token ${token} on css: ${input}`
  );
};

var tokenizer = new CssTokenizer({errorHandler: tokenizerErrorHandler});
var tokens = tokenizer.run(/* input css string */);
/* do something with tokenizer.errors */

and for parser:

const {Parser: CssParser} = parseCss;

const parserErrorHandler = function (parser, input, index, token) {
  parser.errors = parser.errors || [];
  parser.errors.push(
    `Parser error at token ${index}: ${token}`
  );
};

var parser = new CssParser({errorHandler: parserErrorHandler});
var result = parser.parseAStylesheet(/* input css string */);
/* do something with parser.errors */

Also note that this PR doesn't include a change for README yet. A thorough discussion before applying may be needed, as this is a breaking change. Any suggestion is welcome.