zdtm: use tkill instead of kill in sigaltstack

Each signals should be sent to a proper thread..

Probably it was a reason of a error, which was caught yesterday
+ cat zdtm/live/static/sigaltstack.out
20:57:10.933:  4380:        thread in sas: at 0x6083e0 (size   0x2000 flags 0x1)
20:57:10.934:  4380: Waiting in thread SAS
20:57:11.219:  4380: Thread may leave SAS
20:57:11.219:  4380: Leaving thread SAS
20:57:11.219:  4380:        leader in sas: at 0x6083e0 (size   0x2000 flags 0x1)
20:57:11.219:  4380:             main old: at 0x6063e0 (size   0x2000 flags 0x1)
20:57:11.220:  4380:             main new: at 0x6083e0 (size   0x2000 flags 0x1)
20:57:11.220:  4380:             thrd old: at 0x6083e0 (size   0x2000 flags 0x1)
20:57:11.220:  4380:             thrd new: at 0x6083e0 (size   0x2000 flags 0x1)
20:57:11.220:  4380: FAIL: sigaltstack.c:163: sas not restored (errno = 11 (Resource temporarily unavailable))

https://bugzilla.openvz.org/show_bug.cgi?id=2668



Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrey Vagin <avagin@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>