This avoids false negatives with slow name resolutions

Fixes #1636

The vmod packages need this to know if they are
installable or not for this Varnish version.

Leaving it enabled could hide panics.

Fixes #1639

Conflicts:
	bin/varnishtest/tests/v00026.vtc

Conflicts:
	bin/varnishtest/tests/m00018.vtc

This check never worked as intended for several reasons:

- Didn't take the existing file size into account, causing it to
  succeed on first startup but fail on next

- Sparse files were not handled, and can't be handled as we don't know
  where in the file a hole could be and whether truncation actually
  would change the number of blocks used at all

Fixes: #1343

Improve consistency, add argument names and rewording.

Conflicts:
	lib/libvmod_std/vmod.vcc

Do not advance the subject but use the ovector information as offset,
pcre might need to peek back when handling lookbehinds.

Original patch from MegaMaddin via github.

Fixes: #1557

per object, in line with the functionality as it was in Varnish 3.0.

Conflicts:
	bin/varnishd/cache/cache_vrt_var.c

Initial patch by Meng Zhnag - jammy.linux@gmail.com

Conflicts:
	lib/libvmod_std/vmod_std.c

backslash is used for escaping in varnish-cli, so escape the backslash
to actually pass "\.png$".

Minor cosmetic changes while here. Discussed with Dag on Frankfurt.

Move explanation to its own section. Also mention the state options.

This had the same effect as writing to beresp.uncacheble.

Conflicts:
	bin/varnishd/cache/cache_vrt_var.c

With this is possible to find out whether the request was a pass (or
hit-for-pass) in a single place.

Because -0 == +0, the use of negative numbers as a distinction between
hits and misses in the recorded bucket history fails when there are
entries of index 0.

Fixes: #1623

The varnish-counters man page looked so sad for the
counters without long descriptions. Not there yet, but a step
on the way.

In the long description we have sufficient room to
use the full word.