Remove the remains of query categories

[Julian-Wollersberger]

Back in October 2020 in #77830 @cjgillot removed the query categories information from the profiler, but the actual definitions which query was in which category remained, although unused.
Here I clean that up, to simplify the query definitions even further.

It's unfortunate that this loses all the context for git blame, but I'm working on moving those query definitions into rustc_query_system, which will lose that context anyway. EDIT: Might not work out.

The functional changes are in the first commit. The second one only changes the indentation.

[rust-highfive]

r? @davidtwco

(rust-highfive has picked a reviewer for you, use r? to override)

[cjgillot]

A bit of context: I originally kept the query category syntax for a future attempt to split the monolithic query system into per-group subsystems. I since changed my mind, and did not get around to cleaning this up. Thanks @Julian-Wollersberger.

About the blame information: git is able to reconstruct the blame information when performing a git mv. If you are lucky, you won't lose anything when moving all this. We could chose to keep the indentation as-is in rustc_middle::query if this blame information is valuable.

About moving the query declarations to rustc_query_system: how will it interact with #70951? (I still intend to fix the perf regression and get that PR merged some day).

[Julian-Wollersberger]

You're welcome :)

About the blame information

I tried that once, but it doesn't seem to work with GitHub.

About moving the query declarations to rustc_query_system: how will it interact with #70951? (I still intend to fix the perf regression and get that PR merged some day).

I believe it wouldn't have any impact, just that the macro_rules! produced by the query proc macro must be imported from a different path. My plan is to move just the query definitions and the DepKind enum. The rest depends to much on rustc_middle.
That might enable it that a bunch of generic bounds on the DepKind trait could be removed, but I've tried around quite a bit and it seems to not work out thanks to DepKind::debug_node(). So I'll probably have to give up on that.

The original goal was to prevent the query system from being monomorphized for every single query kind. Maybe I can find a different way. Any ideas?

[jyn514]

I tried that once, but it doesn't seem to work with GitHub.

It doesn't work with github, but it works fine locally with git log --follow (which I think is more useful than the online view, which can't do regex searches or things like that).

[davidtwco]

@bors r+

[bors]

📌 Commit 988d93c has been approved by davidtwco