This commit cleans up a number of weeds and loose ends:

The complete_function should not muck about with the pipelining or WS
reservation.

Only when we have dissected the headers for good do we know how long
the headers were.

Make the HTTP1_Dissect* functions have consistent parameters.

before an ESI include at one level, and then again at a higher level.

Fixed by not closing the session in this case, as that will be taken
care of at the end of delivery anyways when V1L_FlushRelease is
called.

Fixes: 		#1737
Submitted by:	Martin

stuff on this TCP connection with timeouts" function.

Inpired by colliding patches from Martin & Lasse.

error has been long retired.

Mark optional parts as such and briefly document components. Sync with
the man page. The address part for options -a and -b is technically
optional too.

We should perhaps rename this parameter away from sigsegv_handler?
Maybe error_sighandler? Maybe move it into a feature bit for the
disabling instead of a separate parameter too.

This allows for a panic dump and backtrace to be produced on
sigabrt. This is useful e.g. when using jemalloc's debug flags, which
can through a special environment variable enable memory allocation
consistency checks and trigger sigabrt on errors.

Reported and diff submitted by Gauthier Delacroix on varnish-dev.

Makes this work with multiple instances under systemd.
Discussed with ingvar on irc.

Fixes startup if vcl_dir was specified in DAEMON_OPTS.
Reported by Gauthier Delacroix. Discussed with ssm on irc.

Fixes #1765

understand them.

Fixes: #1770

For continuous updates we can always use a shell loop.

Fixes #1760.

Please note that using underscore in HTTP headers is considered a
really bad idea because many application frameworks map minus to
underscore in environment variables.

Fixes: 	#1768

Discussed with Martin.