Polyfill finding revision

Hi all,

I've updated the polyfil finding, following our recent call and comments
from community.  Are we able to finalise this in a call soon?

https://docs.google.com/document/d/1u9VgjkPFBgaZE_4xeNCqgF-YReedkTfgXn1WRwmdGFU/edit#

Notable changes:

- Added createShadowRoot example suggested by Domenic
- Removed reference to library authors in introduction to the vendor
prefixing issue, following feedback from Domenic that the prefix problem is
not an issue for polyfills (but it is a good example of naming persistence)
- Changed the definition of the tipping point
- More strongly endorse the term 'speculative polyfill'
- Added advice on naming strategy for polyfills of prototype methods
- Added advice to avoid very generic or short names like $
- Added info about 'partial polyfilling' where a polyfill will supplement
or patch native fucntionality
- Updated list of 'signs of quality'.  This was going to be a list of very
specific points like size, use of timers, rendering efficiency etc, but I
think the document is long enough already so have kept this high level.
- Provided more context to the suggestion that site authors could advise
users that their browser is too old

It's all quite well presented in the diff view if you do File > View
revision history.  The edits are the 1 December entry (for me anyway, with
time zones YMMV)

Finally, there is a significant point of disagreement with Domenic, who has
a strong opinion that polyfills should avoid deferring to native
implementations.  I urged him to raise this on the list, to avoid trying to
have a sizable debate via Google docs comments and to expose the issue to a
wider audience.  This is quite a fundamental point and will probably
benefit from another discussion before we finalise the document.

Cheers,

Andrew