Remove duplicate Transfer-Encoding header

[rht]

Address #2190, #1765.

It is a duplicate since chunked Transfer-Encoding header is already set in http responsewriter.

[rht]

• ipfs cat err, which expects chunked transfer encoding even with the content length.

[jbenet]

cc @dignifiedquire @whyrusleeping @diasdavid who know most about the encodings stuff

[daviddias]

Sweet, @RichardLitt had detected this issue during the HTTP-API docs work :)

@rht what do you exactly mean by ipfs cat err, which expects chunked transfer encoding even with the content length.

[rht]

@diasdavid it's from the travis-ci test: https://travis-ci.org/ipfs/go-ipfs/jobs/103911815#L1458.
I figure that the chunked transfer encoding header doesn't get automatically added due to the request's content-length being explicitly stated.

[rht]

If this line is removed, then the test is fixed. Should it be removed to make for chunk transfer for every response?

[whyrusleeping]

@rht we discussed adding an X-Content-Length header for things like cat that like to know the size of the transfer beforehand, but still want to use chunked encoding

[rht]

"Messages MUST NOT include both a Content-Length header field and a
non-identity transfer-coding. If the message does include a non-
identity transfer-coding, the Content-Length MUST be ignored." [1]

ref:
[1] https://tools.ietf.org/html/rfc2616#section-4.4

[whyrusleeping]

right, which is why i suggested using X-Content-Length instead. so our code can still get the content length for its progress bar stuff

[rht]

I see what you mean. Fixed. travis-ci test fail is unrelated.

[whyrusleeping]

fixed in #2465