Don't accidentally detect the commit hash as an fadd instruction

[Zalathar]

I've seen some reports of tests/codegen/target-feature-inline-closure.rs spuriously failing because it thinks the hash in the rustc version number contains an fadd instruction.

#116085 (comment)
https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/topic/Is.20.60tests.2Fcodegen.2Ftarget-feature-inline-closure.2Ers.60.20flakey
https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/topic/Strange.20.5Cn.20in.20output.20of.20assert.20.23108341/near/395811335

This PR tries to make that not happen by adding a CHECK-LABEL directive that will match the line with the rustc version string, preventing the previous CHECK-NOT from seeing it.

[rustbot]

r? @Mark-Simulacrum

(rustbot has picked a reviewer for you, use r? to override)

[Zalathar]

There may be a more principled or universal way to achieve this, so I'm open to suggestions.

It's a little hard to properly test that this change actually works (especially since I'm not even on x86_64), but locally I did rig up another test to deliberately fail in a similar way, and this does seem to do the trick.

[Mark-Simulacrum]

@bors r+ rollup

I'm happy to land this.

[bors]

📌 Commit 58d62fc has been approved by Mark-Simulacrum

It is now in the queue for this repository.