Skip headers from jupyter kernelspec list.

[ydl7]

On certain configurations, juypter kernelspec list has a header with irrelevant information. This causes an error when parsing JSON output and can simply be ignored.

Example header:

0.00s - Debugger warning: It seems that frozen modules are being used, which may
0.00s - make the debugger miss breakpoints. Please pass -Xfrozen_modules=off
0.00s - to python to disable frozen modules.
0.00s - Note: Debugging will proceed. Set PYDEVD_DISABLE_FILE_VALIDATION=1 to disable this validation.

[nnicandro]

Do you still get those messages even when applying the fix in #418?

[lesteve]

I get this warning with Python 3.11. The fix in #418 (adding --log-level ERROR) does not help:

❯ jupyter kernelspec list --json --log-level ERROR
0.00s - Debugger warning: It seems that frozen modules are being used, which may
0.00s - make the debugger miss breakpoints. Please pass -Xfrozen_modules=off
0.00s - to python to disable frozen modules.
0.00s - Note: Debugging will proceed. Set PYDEVD_DISABLE_FILE_VALIDATION=1 to disable this validation.
{
  "kernelspecs": {
    "python3": {
      "resource_dir": "/home/lesteve/.local/share/jupyter/kernels/python3",
      "spec": {
        "argv": [
          "python",
          "-m",
          "ipykernel_launcher",
          "-f",
          "{connection_file}"
        ],
        "env": {},
        "display_name": "Python 3 (ipykernel)",
        "language": "python",
        "interrupt_mode": "signal",
        "metadata": {
          "debugger": true
        }
      }
    }
  }
}

It seems to me that the best fix is to only consider stdout when executing jupyter commands as was proposed in #388 (which was closed in favour of #418), rather than adding more and more logic to discard unuseful info, which is what #418 did and and this PR proposes.

A compromise could be to add an additional argument to jupyter-command that would allow to ignore stderr. For a command like jupyter kernelspec list --json you know that you need to ignore stderr since otherwise it might not be valid json.