Performance issue for x test tidy

[chenyukang]

x test tidy now cost about 19s on my machine (a M2 mac), I think it's a little bit too long for a trivial fix like #142871, when I discussed with @jieyouxu , he told me it only cost 9s on his dev environment(a WSL).

so I added some logs for cost time in each check in tidy test:
https://github.com/chenyukang/rust/blob/106a6d46375f7c13d4fee3511736193f59b74542/src/tools/tidy/src/main.rs#L87-L95

The result for each part is:

I also meet the issue of #134648, but I do not know how to reproduce it now.

• Do we need to run every tidy check phase for all kind of changes, I guess we may improve on this
• Running them concurrently could speed things up x test tidy can be faster #141074
• I haven't got the reason why it's much slower on my dev compared to @jieyouxu 's dev, maybe network speed ?

[Kobzol]

@GuillaumeGomez This is why I'd rather not run the JS checks by default 😅

[GuillaumeGomez]

We agree on that. I just want it to run by default if a JS file was modified. :)

[petrochenkov]

Only 19 seconds? :D

we@DESKTOP-5ALPNIQ MINGW64 ~/rust
$ time ./x t --bless tidy
Building bootstrap
    Finished `dev` profile [unoptimized] target(s) in 0.26s
Building stage0 tool tidy (x86_64-pc-windows-gnu)
    Finished `release` profile [optimized + debuginfo] target(s) in 0.68s
fmt check
fmt: formatted 14 modified files
tidy check
`eslint` doesn't seem to be installed. Skipping tidy check for JS files.
You can install it using `npm install eslint@8.6.0`
Checking tidy rustdoc_json...
`rustdoc-json-types` was not modified.
No error code explanation was removed!
x.py completions check
Build completed successfully in 0:00:39

real    0m39.629s
user    0m0.060s 
sys     0m0.166s

[RalfJung]

./x test tidy takes 2s here (with a hot build cache)... and that's a laptop, not some beefy workstation. Specifically, this is on Linux with an i7-12700H.

[the8472]

there's also some discussion happening in #142924