Remove dedicated TextCodecUTF16

Remove dedicated TextCodecUTF16

Just rely on ICU for UTF-16LE/BE. This fixes some bugs related to
state handling when decoding chunks with an odd number of bytes. It
also removes non-standard special cases for end-of-stream handling.

BUG=468927, 368904

[jsbell, 2015-03-19 20:58:18]

jsbell@chromium.org changed reviewers:
+ jshin@chromium.org

[jsbell, 2015-03-19 20:58:18]

jshin@ - I don't think we should land this as-is, and I think the removal of
special casing needs more consideration, but for consideration...

[jungshik at Google, 2015-03-20 17:54:14]

On 2015/03/19 20:58:18, jsbell wrote:
> jshin@ - I don't think we should land this as-is, and I think the removal of
> special casing needs more consideration, but for consideration...

Thank you for the CL. I'll play with and take a look at it. 

This (the fact that we have our own UTF-16 converter, which I overlooked) may
explain why a couple of layout tests I made (not uploaded anywhere) to test "not
supporting UTF-32 any more per the encoding spec" fails in an
interesting/unexpected way.

[jsbell, 2015-09-23 17:32:23]

migrated to https://codereview.chromium.org/1366433005/