New comment by tornaria on void-packages repository

https://github.com/void-linux/void-packages/pull/34469#issuecomment-991778989

Comment:
> ```
> aarch64-linux-gnu-g++: fatal error: cannot read spec file 'libgomp.spec': No such file or directory
> compilation terminated.
> ```
> 
> Your cross builds might be hit by #31055

Hum... what shall I do?

The results of the CI are kind of weird; the one for x86_64 finished ok the first time and now it failed. Others failed because a download error (?). I worked hard to fix `do_check()` in four packages so this PR would work :thinking: 

We discussed with @leahneukirchen on IRC how to setup CI so that it is possible to run xbps-src pkg without -Q on cases like this one:

```
10:31 < leah2> gt7: unfortunately not. i always wanted two actions, first build then 
               check, but it seems not possible to implement efficiently?
10:43 < gt7> leah2: maybe an option like [ci-skip] but [check-skip] which will still 
             run ./xbps-src pkg but without -Q ?
10:43 < leah2> yeah, could work
10:45 < gt7> I don't know anything about how github actions work, but the line
10:45 < gt7> TEST: '${{ matrix.config.test }}'
10:46 < gt7> if that can be made conditional to finding [check skip] on the PR title 
             or body, that may work
10:47 < gt7> it should be enough to have an expression that normally evaluates to 
             matrix.config.test, but yields 0 if [check skip] is present in the PR
10:48 -void-robot:#xbps- leahneukirchen pushed to void-packages (stress-ng: update to 
          0.13.08.)
10:49 < gt7> in fact looking at common/travis/build.sh, anything different than 1 
             works so maybe "[skip]" is better so that one can see at the matrix that 
             check was skipped by request
10:50 < gt7> I mean: set TEST to "[skip]" instead of 0 when "[check skip]" is 
             pressent in the PR
```

at least it would save time.