pcp updates, more pdubuf+misc qa

["Frank Ch. Eigler" <fche@xxxxxxxxxx>]

Hi -

For your consideration, presently situated on pcpfans.git fche/multithread:

commit 54a19ad627c4bc4b8fbecc95c0eec2d64384980b (HEAD, origin/fche/multithread, 
fche/multithread
)
Author: Frank Ch. Eigler <fche@xxxxxxxxxx>
Date:   Tue Mar 10 11:19:06 2015 -0400

    qa/common.check: teach _host_to_fqdn() to tolerate no domainname
    
    There exist perfectly functional machines that have a non-fqdn
    hostname(1), an empty domainname(1), and don't happen to show up in
    DNS.  These may be fully reachable on the network via avahi or
    getent/nsswitch/hosts, but several pcpqa tests (e.g., 651) reject
    them.  But alas, even a _notrun would be better than what happens
    currently:
    
    This patch rejects the _host_to_fqdn() heuristic's attempt to suffix
    the incoming hostname with `domainname`, if that result was "(none)".
    Passing "(none)" back as part of a fqdn is a recipe for tears and qa
    failure.

commit f8a9722fb7274dd4be7beb72864c38c3ea702bf8
Author: Frank Ch. Eigler <fche@xxxxxxxxxx>
Date:   Tue Mar 10 09:55:28 2015 -0400

    pdubuf padding: initialize in __pmSendLogStatus
    
    The logstatus_t.pad field needs to be initialized to prevent
    information leakage, as detected by valgrind.

commit b8f7582da2003d1454015a32aeb16ea2eab04475
Author: Frank Ch. Eigler <fche@xxxxxxxxxx>
Date:   Mon Mar 9 12:53:42 2015 -0400

    qa/063: tweak embedded C test case to satisfy gcc5
    
    gcc5 emits:
    
    warning: return type defaults to 'int' [-Wimplicit-int]
     main() {
     ^
    
    so let's plop an 'int' there.