Optionally support a parameter object to control internal service shards

See merge request uplex/varnish/k8s-ingress!8

k8s uses re2, which does not support lookaround, so this rather
complicated pattern is what I came up with to forbid vcl and
vk8s prefixes.

The shard director supports many interesting use cases, for example

- retrying backend requests on different backends

- sharding by custom keys, for example for "client stickyness"
  or cache optimization

This patch adds the option to create parameter objects for backends
using the shard director.

Each named parameter object will be created only once and associated
with all directors for which the backend configuration names them.

Sleep a few seconds longer before grepping the controller log.
During a lengthy e2e test, the expected log line was not found
by grep in time.

The default format has been changed to include the instance name
and the test only needs to check for a sensible Via header at all.

... a COPYING file exists.

See
* https://github.com/libsndfile/libsndfile/pull/819
* https://github.com/libsndfile/libsndfile/pull/820

E: Version '2.2.9-2+deb11u2' for 'haproxy' was not found

We had updated to a very new version from bullseye-backports to test
if it had a necessary bugfix, but that brought no improvement. As it
turned out, the problem was ulimit -n as addressed in 03f46509.

2.2.9 is sufficient for our purposes. backports repos are problematic
because a version of a package ceases to exist from time to time.
This causes container builds to fail at times that are unpredictable
in practice.

The field is still optional, but if no default key is specified, then
it is implicitly the empty string. We no longer support a use case
in which an error response is returned for sharding by key if the
key is missing from the request. This will also be the case when
we support sharding by other keys besides cookies, such as request
headers.

This makes it possible to move the clause in vcl_recv that includes
the synthetic response to health checks from other shard instances
to come after the code for primaryOnly. That had been moved to the
start of vcl_recv, because health checks were failing in the shard
by cookie case without a default key (since health probes have no
Cookie header). But the result was that primaryOnly code did not
work as expected. Now we have moved that clause back to where it
was previously -- health probes now don't fail in the shard by
cookie case, because there is always a default key.