Update generated package.jsons

[lishaduck]

Trying to make the generated package.jsons more accurate.

Extracted from #303.

[lishaduck]

Pros:

• Direct call (perf)
• Clearer dependency chain (sorta con, see below)
• Works even if npx isn't on path
• Fewer Windows gotchas

Cons:

• Would show up as unmaintained on socket, etc (which is true, it is)
• Depends on deprecated packages (which is true, we do)
• Bloats node_modules for 99% of users

[jfmengels]

We could also inline the functionality of the package: support only the common ones and let the user opt-out if they want a different one.

[lishaduck]

I've noticed this there's a test that's been flaking ("Running using other configuration (without errors)"):

-- UNEXPECTED ERROR ------------------------------------------------------------

I ran into an unexpected error. Please open an issue at the following link:
  https://github.com/jfmengels/node-elm-review/issues/new

Please include this error message and as much detail as you can provide. Running
with --debug might give additional information. If you can, please provide a
setup that makes it easy to reproduce the error. That will make it much easier
to fix the issue.

Below is the error that was encountered.
--------------------------------------------------------------------------------
<local-path>/test/project-with-errors/elm-stuff/generated-code/jfmengels/elm-review/cli/<version>/review-applications/bd2a2bb6c965dac5241ce8c4d0615d83.js:10328
    		$temp$offset = offset + 
    						                    

SyntaxError: Unexpected end of input

  at loadCompiledElmApp (<local-path>/lib/load-compiled-app.js:28:18)
  at initWithoutWorker (<local-path>/lib/app-wrapper.js:144:21)
  "
  at Object.test (test/review.test.js:37:1)

[jfmengels]

I've noticed this there's a test that's been flaking ("Running using other configuration (without errors)"):

Yes, I've noticed the same thing. I had a look, thinking it could be us not waiting for the compilation to be finalized, but the code looks fine. I still think it's the writing of the file (or the loading of it, but that's weirder) that is getting cut off early somehow, but I haven't figured out how yet. 🤔

It sounds pretty recent though. I don't think we saw this one before, but I could be wrong.

[jfmengels]

Thank you for the changes!

[lishaduck]

It sounds pretty recent though. I don't think we saw this one before, but I could be wrong.

Well, it's really started flaking. Will look into it.
For reference, here's the first failure from it: https://github.com/jfmengels/node-elm-review/actions/runs/11769034390/job/32779541170

Commit-wise, there was a different flake on #306 (it's older than my work here, I think I remember seeing it a while ago), and then fs-extra (#303) seems to have started this flake.

Am gonna try to see if I reproduce it before #303.

[lishaduck]

Was able to repro with 38453c1, so sadly not #303.

[lishaduck]

Huh. I can't repro on #304 for the life of me,¹ so I guess it's gotta be #306 (or, I suppose, #305), which makes no sense.

Footnotes

1. I've run tests 10+ 9 times on the commit, to no avail. ↩

[lishaduck]

@jfmengels, any ideas?