Allow responses of HEAD requests to have empty DATA frames

[eaufavor]

Per https://tools.ietf.org/html/rfc7540#section-8.1.2.6

A response that is defined to have no payload, as described in [RFC7230], Section 3.3.2, can have a non-zero content-length header field, even though no content is included in DATA frames.

Without this patch, h2 errors with recv_data: content-length overflow; stream=StreamId(1); len=0; when empty DATA frame is received.

[seanmonstar]

Nice catch, thanks!

Do you think we could add a test to tests/client_request.rs so we don't regress this?

[eaufavor]

Added a test.

[eaufavor]

The h2spec test might have some troubles in CI. It has been hanging for hours since my first commit.
Locally it passes quickly.

Finished in 0.3034 seconds
145 tests, 144 passed, 1 skipped, 0 failed
h2spec passed!

[seanmonstar]

Hm, how weird! I can't even get it to show me the output so far, to try to understand what it is hanging on...

[eaufavor]

It will be killed after hanging 6 hours.
https://github.com/hyperium/h2/runs/1249393890 You can see the log via the "Download log achieve"

2020-10-12T23:37:39.9002362Z   3100K .......... .......... ........                        100% 93.3M=0.07s
2020-10-12T23:37:39.9002599Z 
2020-10-12T23:37:39.9004454Z 2020-10-12 23:37:39 (41.9 MB/s) - ‘h2spec_linux_amd64.tar.gz’ saved [3203286/3203286]
2020-10-12T23:37:39.9004788Z 
2020-10-12T23:37:40.0703625Z     Finished dev [unoptimized + debuginfo] target(s) in 0.05s
2020-10-13T05:33:19.5875096Z ##[error]The operation was canceled.
2020-10-13T05:33:19.5902345Z Cleaning up orphan processes
2020-10-13T05:33:19.6077171Z Terminate orphan process: pid (12380) (h2spec.sh)
2020-10-13T05:33:19.6102023Z Terminate orphan process: pid (12472) (h2spec.sh)
2020-10-13T05:33:19.6121418Z Terminate orphan process: pid (12473) (sed)
2020-10-13T05:33:19.6141672Z Terminate orphan process: pid (12474) (server)

So it looks like it hangs on starting the server.