Cuirass remote worker is stuck building with 1 core

Cuirass remote worker is stuck building with 1 core

[Parnikkapore]

I noticed that the Cuirass remote worker in my channel-building VM only builds things with 1 core, despite me having configured Guix to use multiple cores per build.

# Reproduction instructions

On a multi-core QEMU VM running Guix System with a Cuirass setup, run the following command:

```
cuirass remote-worker -w 1
```
This will result in the message "starting 1 workers (parallelism: 1 cores)".

# Expected output

The remote worker tries to build with the number of cores specified in guix-daemon's --cores. That, or $(nproc).

It should be noted that my bare-metal foreign distro install correctly selects $(nproc)-core parallelism. However, I have created multiple QEMU VMs running Guix System, and they all reproduce this bug.

# Appendix

## channels.scm

```
(list (channel
        (name 'guix)
        (url "https://git.savannah.gnu.org/git/guix.git")
        (branch "master")
        (commit
          "577fd21122ca50ab7594be9adf0c5c06d5b26678")
        (introduction
          (make-channel-introduction
            "9edb3f66fd807b096b48283debdcddccfea34bad"
            (openpgp-fingerprint
              "BBB0 2DDF 2CEA F6A8 0D1D  E643 A2A0 6DF2 A33A 54FA")))))
```

## repro script

```
cd "$(dirname "$0")" | exit
guix time-machine -C ./channels.scm -- shell cuirass -- sh -c 'sudo $(which cuirass) remote-server --no-publish -c /tmp/cuirass-cache & sudo $(which cuirass) remote-worker -w 1'
```

Re: Cuirass remote worker is stuck building with 1 core

[Parnikkapore]

I investigated this a bit more:
* issue is reproduced with <=4 cpus on the VM; apparently it does not occur if there are more cpus
* (current-cpu-count) is 1 within the worker thunk on affected setups, but it's not affected in cuirass-remote-worker