feat(cli): Add silent / quiet global CLI flag

[paulomarg]

Closes #23192

As mentioned in the issue, there are some scenarios where we want to omit the regular message returned from a command, where the output might make it harder to spot legitimate issues, or bloats the output of some scripts.

To achieve that, this PR introduces a new global --quiet flag (aliased to --silent if I understood the setup correctly 😄) that simply ignores the message returned from a command, unless that is an Error.

With this, users can toggle the regular output off but still catch any errors thrown by commands so we don't accidentally hide problems. Messages logged directly by a command (assuming no normal flow messages are logged that way) are still output normally.

[CLAassistant]

All committers have signed the CLA.

[codspeed-hq]

CodSpeed Performance Report

Merging #24232 will improve performances by 22.52%

_{Comparing paulomarg:add_silent_flag_to_cli (0ec450a) with paulomarg:add_silent_flag_to_cli (c9f0a71)}

Summary

⚡ 1 improvements
✅ 2 untouched benchmarks

Benchmarks breakdown

	Benchmark	paulomarg:add_silent_flag_to_cli	paulomarg:add_silent_flag_to_cli	Change
⚡	client generation 100 models with relations	8 s	6.6 s	+22.52%

[SevInf]

Hi @paulomarg
Thank you for submitting the PR. --quiet flag makes sense, however, I am pretty sure this PR won't quite achieve the functionality you want - unfortunately, we are pretty bad at centralizing our logging and there are bunch of other places within the CLI where rogue console.log might slip in.
If you want to do an attempt at centralized logging, we'll take it. However, that might be a lot of work and I think there is way for us to proceed with this PR - just that before we can merge it, we should at least be sure that we won't make this logging situation worse in the future. And for that we'd need to snapshot output from real commands, not the fake one. You can find the examples of how to do it in cli/src/__tests__ and migrate/src/__tests__.

Additionally, small nitpick - I don't think there is a good reason to have both --silent and --quiet flag, I'd prefer we'd just keep --quiet for now.

Thank you again for this PR and let me know if you need any help.

[paulomarg]

Hey, thanks for getting back to me! That was a good callout, I added some test that would check the actual output and it does seem like a lot of it isn't coming from the command return value (see the new snapshots), so my proposal was too shallow.

I agree that the right solution would have to be centralizing all of the logging and using the flag to turn that off instead, but unfortunately I won't have the time to do that :(

Thanks for taking the time to review, and I hope this PR helps whatever solution you end up implementing! I'm going to close it since it doesn't feel like it'll solve the problem.