Initialize struct with zeroes

[MrZammler]

Summary

Zero initialize write_ctx to remove a valgrind warning in spawn/spawn_client.c.

This should remove this warning:

==7869== Thread 2:
==7869== Syscall param writev(vector[...]) points to uninitialised byte(s)
==7869==    at 0x55D3C1D: writev (in /lib64/libc-2.33.so)
==7869==    by 0x4E29909: ??? (in /usr/lib64/libuv.so.1.0.0)
==7869==    by 0x4E29AD8: ??? (in /usr/lib64/libuv.so.1.0.0)
==7869==    by 0x4E2AC25: uv_write2 (in /usr/lib64/libuv.so.1.0.0)
==7869==    by 0x28E641: spawn_process_cmd (spawn_client.c:160)
==7869==    by 0x28E641: spawn_client (spawn_client.c:215)
==7869==    by 0x54C8E5D: start_thread (in /lib64/libpthread-2.33.so)
==7869==    by 0x55DCEFE: clone (in /lib64/libc-2.33.so)
==7869==  Address 0x5a67d54 is 196 bytes inside a block of size 240 alloc'd
==7869==    at 0x483F855: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==7869==    by 0x15A64B: mallocz (libnetdata.c:177)
==7869==    by 0x28E591: spawn_process_cmd (spawn_client.c:142)
==7869==    by 0x28E591: spawn_client (spawn_client.c:215)
==7869==    by 0x54C8E5D: start_thread (in /lib64/libpthread-2.33.so)
==7869==    by 0x55DCEFE: clone (in /lib64/libc-2.33.so)
==7869==  Uninitialised value was created by a heap allocation
==7869==    at 0x483F855: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==7869==    by 0x15A64B: mallocz (libnetdata.c:177)
==7869==    by 0x28E591: spawn_process_cmd (spawn_client.c:142)
==7869==    by 0x28E591: spawn_client (spawn_client.c:215)
==7869==    by 0x54C8E5D: start_thread (in /lib64/libpthread-2.33.so)
==7869==    by 0x55DCEFE: clone (in /lib64/libc-2.33.so)
==7869== 

Component Name

Spawn server

Test Plan

Since spawn server is mainly used for running the alarm-notify.sh script, it should be tested that after this PR, notifications still get executed, but there is no warning from valgrind for this case.

Additional Information

[thiagoftsm]

Using the following alert:

 alarm: dev_dim_template
    on: system.cpu
    os: linux
lookup: sum -3s at 0 every 3 percentage foreach *
 units: %
 every: 1s
  warn: $this > 1
  crit: $this > 4

and running netdata inside valgrind-3.17.0 I did not observe more the reported problem.

I also did not observe any warning with libasan. For both scenarios I waited for 100 alerts.

LGTM!

[MrZammler]

Thanks a lot!!