resolv.conf.5: Timeout does not map to resolver API calls

resolv.conf.5: Timeout does not map to resolver API calls

I'm posting this patch to clarify the timeout behaviour because
there have been developers who expect this timeout to mean
something it is not.

The timeout (and by proxy attempts) does not map to resolver API
calls.  For example a single call to getent might involve multiple
resolution requests to the resolvers listed in resolv.conf and
each request will use TIMEOUT and be attempted at least ATTEMPT
times. A developer using the resolver API cannot easily compute
any given timeout because the implementation may change e.g. A and
AAAA queries made in parallel.  A system administrator uses this
setting to ensure there is a desirable timeout on any request to
any of the nameservers listed in resolv.conf, but no guarantees
exist beyond that.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
Signed-off-by: Carlos O'Donell <carlos@redhat.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>