feat: implement service worker example

[Gozala]

Adds example demonstrating how to use IPFS node in shared worker from service worker.

[achingbrain]

It looks like the test for the new example-browser-service-worker example is failing?

[achingbrain]

There's a webpack error before the test runs:

example-browser-service-worker: [webpack-cli] [Error: ENOTDIR: not a directory, open '/home/travis/build/ipfs/js-ipfs/examples/browser-service-worker/dist/main.js'] {
example-browser-service-worker:   errno: -20,
example-browser-service-worker:   code: 'ENOTDIR',
example-browser-service-worker:   syscall: 'open',
example-browser-service-worker:   path: '/home/travis/build/ipfs/js-ipfs/examples/browser-service-worker/dist/main.js'
example-browser-service-worker: }

[Gozala]

@achingbrain fixed it all passes now, was just a typo in test scripts where I had & instead of &&.

[achingbrain]

There's an awful lot of boilerplate code here, it's not terribly clear what the minimal set of code needed to get a shared worker going is and what is window dressing.

It also uses a lot of technology that may not be absolutely essential and as such are distractions. People look at these examples without a 100% understanding of TypeScript, Webpack, Babel, even JavaScript so anything that can be removed should be removed.

E.g. People will largely copy/paste the example and use it to start a new project. Ask yourself what tech choices have you then made for them? Maybe don't make those choices. For example if we can get away with just using a script tag and linking to node_modules/some-module/dist/index.min.js instead of using a bundler, do that instead - it'll make the example more concise (good for the reader) and more maintainable since there will be fewer dependencies to be constantly upgrading (good for us).

Could you please restructure the example to show clearly what is boilerplate/unnecessary and what is not.

[Gozala]

@achingbrain as per our discussion I have moved some pieces into utils and added a general overview section into readme that attempts to provide guidence in terms of how to navigate the code base.

[Gozala]

@achingbrain I have addressed remaining notes

• instead of error base dir rendering fallback callback now code uses ipfs.file.stat to choose between dir & file renderring.
  • There are still some error handling e.g. to redirect non file / dir entries to https://explore.ipld.io for the nodes
• dir renderring no longer buffers but rather streams response.
  • Both chrome and firefox seem to do incremental rendering of entries
  • I have added limit to the number of entries, if limit is reached note about pagination user exercise is renderred