I was in the neighborhood...

This reverts commit 6233b088.

Conflicts:
	bin/varnishd/cache/cache_req.c

The conflict was introduced with the handling of h2 stream 0 pseudo
request: 51127b46. This fixes a
regression where timeout_idle could be circumvented by a periodic
CRLF, except that it has no real consequence like holding onto a
worker thread thanks to timeout_linger kicking in as expected.

In addition to reverting the change, test coverage is added.

This reverts commit 762fda9a.

Failure to read the right number of bytes (typically due to remote HUP)
would log "straight insufficient bytes", which is the same as we log for
C-L based fetches. Change this to "chunked insufficient bytes". (Assuming
this was a copy-paste error).

Ref #3088

apparently the additional argument was just silently tolerated by
varnishtest

Tests #3088

Using vut->last_sighup was not adequate, because the VUT_Main loop might
have terminated for some other signal or because of EOF.

On the other hand, we do not want to end the curses loop just because
the VUT_Main loop has ended in order to continue to display the last
state (and maybe the EOF in the top right corner).

So while I see that checking just one flag has some beauty to it, I do
think that directly checking the signal counters is both simple and
robust.

Fixes #3088 for varnishtop and varnishhist

Fixes #3088 for varnishstat

note that besides handling SIGHUP, this also adds handling of SIGTERM
and SIGINT, which were also missing.

Refs a791188e

backend None does not actually create a backend, so tests which require
one need to use the bad_ip method.

... for those cases which can easily be changed (most of them).

This avoids issues with failing tests when bad_ip is actually reachable
via tcp on the respective port, as seen for example with b00068.vtc

by properly cleaning up the busyobj

Also move the relevant code from cache_vrt.c to cache_fetch.c

As we fini the director during cleanup, we now also need to handle the
backend connection gone missing in vbf_stp_fetch(). The hypothetical
alternative would be to not fini the director, but I believe this is not
safe in case it also used some workspace.

Fixes #3009

in particular because it may live on the workspace.

Required for #3009

imples the following changes:

* VDI_Finish() is now always conditional on bo->director_state !=
  DIR_S_NULL, making it idempotent

* introduces additional calls to VFP_Close() from startfetch and
  for the filter_list / VCL_StackVFP error in vbf_stp_fetch(),
  but VFP_Close() is idempotent.

* adds VFP_Close() for VFP_Open() failure in vbf_stp_fetch() which
  I think is actually missing (for the case that some VFPs could
  get opened before the open failure)

* calls VDI_Finish() earlier in vbf_stp_fetchend: I checked the
  code and can not see any issue with this.

motivated by #3009

required for #3009

On a network where you can't trust the DNS resolver because it filters
AAAA records out v00016.vtc fails.

... to avoid future regressions of a very useful feature.

Using a gap in the test numbers.

We should avoid the word "received" in the context of response fields,
since we don't "receive" resp.* fields, only beresp. This also fixes
another inconsistency in the manual:

    RespReason - Client response response
        The HTTP response string received.

Ref #3067

@andrewwiik would you please look after the user documentation?