fix: exec command with failFast should fail immediately

[MohiuddinM]

Description

The expected behavior of failFast option is to fail the whole exec command as soon as a single fail occurs, and it works as expected if concurrency=1. But if concurrency > 1, it diverges from the expected behavior and instead keeps running until all pooled jobs have finished executing.

This PR fixes this divergence, and the exec command will fail immediately on a single fail irrespective of concurrency value.

Type of Change

• [ ] ✨ feat -- New feature (non-breaking change which adds functionality)
• [x] 🛠️ fix -- Bug fix (non-breaking change which fixes an issue)
• [ ] ❌ ! -- Breaking change (fix or feature that would cause existing functionality to change)
• [ ] 🧹 refactor -- Code refactor
• [ ] ✅ ci -- Build configuration change
• [ ] 📝 docs -- Documentation
• [ ] 🗑️ chore -- Chore

[MohiuddinM]

Note to maintainers: only the last commit is related, others are showing because I didn't fork specifically for this fix, and instead merged this into my fork.

[MohiuddinM]

@spydon I have:

• Added additional logger to report canceled tasks separately from failed jobs:

          '$ melos exec\n'
          '  └> exit 1\n'
          '     └> FAILED (in 1 packages)\n'
          '        └> c (with exit code 1)\n'
          '     └> CANCELED (in 2 packages)\n'
          '        └> a (due to failFast)\n'
          '        └> b (due to failFast)\n'

• Added tests
• Refactored earlier implementation to remove exit (to make code testable)

[MohiuddinM]

Tests are flaky I guess because all 3 are being run at the same time, and some times "a" fails first, and "b", "c" are cancelled. Otherwise, some other fails. I need to find a way to add delay before "exit 1" so it becomes more predictable.

   Expected: String ignoring Ansii '$ melos exec\n'
              '  └> exit 1\n'
              '     └> RUNNING (in 3 packages)\n'
              '\n'
              '--------------------------------------------------------------------------------\n'
              '--------------------------------------------------------------------------------\n'
              '\n'
              '$ melos exec\n'
              '  └> exit 1\n'
              '     └> FAILED (in 1 packages)\n'
              '        └> c (with exit code 1)\n'
              '     └> CANCELED (in 2 packages)\n'
              '        └> a (due to failFast)\n'
              '        └> b (due to failFast)\n'
              ''
    Actual: '$ melos exec\n'
              '  └> exit 1\n'
              '     └> RUNNING (in 3 packages)\n'
              '\n'
              '--------------------------------------------------------------------------------\n'
              '--------------------------------------------------------------------------------\n'
              '\n'
              '$ melos exec\n'
              '  └> exit 1\n'
              '     └> FAILED (in 1 packages)\n'
              '        └> b (with exit code 1)\n'
              '     └> CANCELED (in 2 packages)\n'
              '        └> a (due to failFast)\n'
              '        └> c (due to failFast)\n'

[MohiuddinM]

@spydon I found a solution to the random failure order by adding delays (see: https://github.com/invertase/melos/pull/665/commits/1ab3177051b704a199f7db8b1e86e91acc0c9c08).

The test failing now is unrelated I don't know why, but it passes on my PC.