feat(log): unhide tracing::instrument from behind feature = "otel"

[rami3l]

Cherry-picked from #3803 as discussed in #3803 (comment).

Note

This is done by simply replacing cfg_attr\(feature = \"otel\", tracing::instrument(\((.*)\))?\)\] with tracing::instrument(level = "trace", $2)], then replacing , \)\] with )].

[rami3l]

The Windows CI is still red, but only a small number of tests are failing due to stack overflow compared to #3803's CI.

@roife has pointed me to the stack size restriction on Windows, saying that maybe it's not a stack overflow caused by infinite recursion, instead we might simply need a bigger stack. rust-analyzer has some code related to this restriction: https://github.com/rust-lang/rust-analyzer/blob/6b8b8ff4c56118ddee6c531cde06add1aad4a6af/crates/rust-analyzer/src/bin/main.rs#L147-L163

[ChrisDenton]

On Windows, applications choose their stack size rather than being defined by the system. So another option is to increase the default stack size by using a linker argument (for msvc that's /stack). I don't know what's easier.

[rami3l]

@ChrisDenton Thanks a lot for your pointers!

Looks like unoptimized Rust is indeed hungry on stack space:

The Rust compiler uses a 6MiB stack and then increase than to 20MiB (!!!) when optimizations are disabled. The default stack size on Linux is 8MiB, so that's likely how much is available for the main thread. Rust tends to need a ridiculous amount of stack space, especially without optimizations.
rust-lang/rust#17044 (comment)

I'll try your suggestion in https://users.rust-lang.org/t/stack-overflow-when-compiling-on-windows-10/50818/8 and rust-lang/rust#85303 first.

[ChrisDenton]

Oh, hm actually I think the test runner sets the stack size itself. I'm not sure if that is directly configurable. You might need to set the RUST_MIN_STACK environment variable.

[rami3l]

Oh, hm actually I think the test runner sets the stack size itself. I'm not sure if that is directly configurable. You might need to set the RUST_MIN_STACK environment variable.

@ChrisDenton Thanks a lot! That seems to work :)

PS: Sorry I got a bit sidetracked with other issues in the past few weeks 🙈