Seems reasonable.

So it looks like this doesn't work as intended.

I made the nightly workflow run on PR in the top most commit.

• nightly
• c-i

Both of those show in their test output:

ok 19 all container contents must be accounted for # skip test_all_container_contents_must_be_accounted_for is slow. Set SLOW_TEST=true to run.
ok 20 bom tool should work inside run # skip test_bom_tool_should_work_inside_run is slow. Set SLOW_TEST=true to run.

Further investigation shows that I had SLOW_TESTS (plural) in the workflow 'env' while the code is looking at SLOW_TEST (singular).

Change made, pushed, lets see.

After this runs i have to still drop the top-most commit so that nightly doesn't run on PR.

both of the slow test runs failed. see here.

• unpriv failed:

  not ok 48 elasticsearch in 177sec
  # (from function `stacker' in file test/helpers.bash, line 87,
  #  in test file test/convert.bats, line 71)
  #   `stacker build -f stacker.yaml --substitute-file stacker-subs.yaml --substitute IMAGE=elasticsearch' failed
  

• priv failed:

  not ok 7 mount + umount works in 49sec
  # (from function `stacker' in file test/helpers.bash, line 87,
  #  from function `basic_test' in file test/atomfs.bats, line 28,
  #  in test file test/atomfs.bats, line 36)
  #   `basic_test' failed
  # Debug mode: 
  

maybe the 'mount + umount' failure is unrelated as it normally passes, but the other failure is in a test marked slow.

I'd like to ask that this be landed. nightly c-i will still be broken, but this PR gets us closer.
Once landed, I think you'll be able to run the nightly manually on a branch to debug/fix it.
See https://docs.github.com/en/actions/using-workflows/manually-running-a-workflow for how to do that.