DisplayFn trait for print fns on clean rustdoc items

[yotamofek]

Not sure if I like this change or not, but it has some pros (and cons) so thought I'd put it up here for discussion.

In #148585 (comment), I lamented the fact that there was a lot of boilerplate around our print_* fns, due to the fmt::from_fn opening a new block (and indenting everything by another tab).
Using a macro to change the signature of a function is definitely not a good idea, so I attempted to do something similar with a trait and a blanket impl.

Basically, the DisplayFn trait is implemented for fns that take a &mut fmt::Formatter, a "clean" item, and a rustdoc ctx, and adds a display_fn method on such functions that does the whole fmt::from_fn wrapping.

Pros:

• De-indent all our print fn code by one level
• Maximalist API - call sites that already have a &mut Formatter can just pass it to the original function, while those that don't can use the display_fn trait method
• IMHO improves readability for both the fns and the call sites

Major con:

• Blanket impls on Fns are not the greatest DX. If you get a function signature wrong, the diagnostics are not going to be very helpful. And it's not really clear where that display_fn method is coming from (though I don't think it should matter much most of the time, since its functionality is pretty easy to infer from the context, IMHO)

r? @GuillaumeGomez feel free to just close this if you don't like the idea, not married to it myself :)

[rust-log-analyzer]

The job aarch64-gnu-llvm-20-1 failed! Check out the build log: (web) (plain enhanced) (plain)

Click to see the possible cause of the failure (guessed by this bot)

---- [rustdoc] tests/rustdoc/return-type-notation.rs stdout ----
------python3 stdout------------------------------

------python3 stderr------------------------------
11: has check failed
 `XPATH PATTERN` did not match
 //@ has - '//pre[@class="rust item-decl"]' "pub fn foo<T: Foo<bar(..): Send>>()"

Encountered 1 errors

------------------------------------------

error: htmldocck failed!
status: exit status: 1
command: "/usr/bin/python3" "/checkout/src/etc/htmldocck.py" "/checkout/obj/build/aarch64-unknown-linux-gnu/test/rustdoc/return-type-notation" "/checkout/tests/rustdoc/return-type-notation.rs"
stdout: none
--- stderr -------------------------------
11: has check failed
 `XPATH PATTERN` did not match
 //@ has - '//pre[@class="rust item-decl"]' "pub fn foo<T: Foo<bar(..): Send>>()"

Encountered 1 errors
------------------------------------------

---- [rustdoc] tests/rustdoc/return-type-notation.rs stdout end ----

[GuillaumeGomez]

I'm personally not very convinced by this approach. ^^'

We basically exchange a level of indent for a level of indirection. It's a neat trick though.

[yotamofek]

I can't convince you of something I don't believe in myself 😬

[GuillaumeGomez]

I mean, we could bring this up at next rustdoc meeting (which is tomorrow haha).