Skip unit-test harnesses when building benchmarks (#9158)

cargo bench and cargo codspeed build select every target with bench = true.
That includes the library and binary targets, so the build also compiled the
library unit-test harness as an extra LTO-linked copy of the interpreter that
no benchmark runs. Mark them bench = false so only the criterion benchmarks
(and the binary they depend on) are built.

Co-authored-by: CodSpeed Bot <no-reply@codspeed.io>

Cargo.toml

@@ -68,6 +68,13 @@ rustls-graviola = { workspace = true }
 rustpython-stdlib = { workspace = true }
 ruff_python_parser = { workspace = true }
 
+# Only the criterion benchmarks below are measured. Without `bench = false`,
+# `cargo bench` (and `cargo codspeed build`) also builds the library and binary
+# unit-test harnesses with the LTO bench profile, adding a full extra link of
+# the interpreter that is never run.
+[lib]
+bench = false
+
 [[bench]]
 name = "execution"
 harness = false
@@ -79,6 +86,7 @@ harness = false
 [[bin]]
 name = "rustpython"
 path = "src/main.rs"
+bench = false
 
 [[example]]
 name = "custom_tls_providers"