Add a proper support for a machine reboot

[metan-ucw]

There are several different test types that cause reboot/crash of the SUT. Examples are reboot(), kexec() syscalls, MCE error injection, kernel crash dump testing, etc. Currently such tests are impossible to implement and many of the features are untested.

To get this working we need:

• add .reeboots_sut flag into the tst_test structure
• kirk has to parse the metadata and expect the SUT to be rebooted after such tests
  • such test cannot run in parallel
  • kirk has to be able to reestablish the connection to SUT after it booted up
• these tests would need additional command line parameter since the test binary will be executed twice
  • first time the test is executed before reboot
  • second time it's executed after reboot
  • these test must just print warning and do not do anything if this parameter is not provided

Example workflow would be:

• kirk executes reboot01 -P1 that calls reboot() syscall to trigger the system reboot
• kirk waits for the machine to shutdown and boot
• once machine is up again kirk executes reboot01 -P2 which would check that system is freshly booted
  (uptime is small enough or something similar)

[metan-ucw]

@tbird20d FYI