fix(agent): make sure phase exit codes are in the 0-255 range

[boukeas]

Description

When a process is terminated due to a "stop condition checker", it ends up receiving a SIGKILL and the returncode for a process that receives a SIGKILL (signal 9) is -9. That returncode is used as the exit status of the phase, which means that exit statuses reported by Testflinger can be negative, rather than in the expected range of 0-255.

This PR ensures that phase exit codes are in the 0-255 range.

Tests

In a local Testflinger deployment, this job was used for testing, deliberately timing out:

job_queue: myqueue
output_timeout: 60
test_data:
  test_cmds: |
    echo "Starting"
    sleep 120
    echo "Ending"

Here are the results when the agent was deployed from main:

{
    "job_state": "complete",
    "test_output": "************************************************\n* Starting testflinger test phase on agent-007 *\n************************************************\n2024-10-21 10:54:54,795 myqueue INFO: DEVICE CONNECTOR: BEGIN testrun\n\nERROR: Output timeout reached! (60s)\n",
    "test_status": -9
}

And here are the results when the agent was deployed from this branch:

{
    "job_state": "complete",
    "test_output": "************************************************\n* Starting testflinger test phase on agent-007 *\n************************************************\n2024-10-21 10:52:13,757 myqueue INFO: DEVICE CONNECTOR: BEGIN testrun\n\nERROR: Output timeout reached! (60s)\n",
    "test_status": 247
}

[plars]

Looks good, but can we add a small unit test for this? Just something to ensure that when we timeout, that it gets the expected (now positive) _status for the phase should suffice.

[plars]

+1, thanks!