Revert "Backport "Always traverse Inlined.call in linter" to 3.7.4"

[WojciechMazur]

Reverts #24227 from Scala 3.7.4-RC3 due to #24266 and other regressions, that are unlikely to be fixed in a required time window (we want to release 3.7.4 stable next week, which leaves at most 1 day for developing and testing the fixes).
Would be kept in the 3.8 series

[som-snytt]

Note that llm4s build OOMs trying to Vprint:inlining, but one would not revert -Vprint on that basis.

[som-snytt]

We're still at 0.5 days, simple PRs are available for review, if desired.

[Gedochao]

@WojciechMazur I guess the alternative would be to back port the 3 fixes from last night (they still need review):

• Lint avoids revisiting Inlined.call #24277
• Register no elements for lint after inlining #24279
• Use enclosing enclosingInlineds for empty call #24281

[som-snytt]

An argument in favor of reverting is that people will be eager to upgrade to 3.8. I think the PRs are an improvement, and small in scope, but it's also nice to have time for feedback. My only sense of urgency is that it's hard to get PRs merged, so that it's better not to step backward; but timing is everything.

[WojciechMazur]

Let's test these 3 branches backported to release-3.7.4 against the OpenCB. If there would be no new issues, let's merged them as it's also fixing the regressions introduced in 3.8.0, otherwise let's revert and delay until 3.8.

[WojciechMazur]

We'd use the backported fixes instead of this revert - all the known issues are fixed, no performance issues visible, no new regressions, scala-cli passes the build (when using main 3 subprojects are failing but it's the same when using 3.7.3 or 3.7.4-RC3)
@som-snytt Thank you for your heroic work!

[som-snytt]

@WojciechMazur Thanks for taking a chance. I'm available all weekend for follow-ups. Probably the world series is over by then.