wc: count characters as bytes in single-byte locales

Cargo.toml

@@ -119,6 +119,7 @@
 rustix = { version = "1.1.4", default-features = false }
 self_cell = "1.0.4"
 selinux = "0.6"
+simdutf8 = "0.1.5"
 tempfile = "3.15.0"
 terminal_size = "0.4.0"
 textwrap = { version = "0.16.1", features = ["terminal_size"] }

src/uu/wc/BENCHMARKING.md

@@ -27,9 +27,11 @@
 
 ### Counting lines and UTF-8 characters
 
-If the flags set are a subset of `-clm` then the input doesn't have to be decoded. The
-input is read in chunks and the `bytecount` crate is used to count the newlines (`-l` flag)
-and/or UTF-8 characters (`-m` flag).
+If the flags set are a subset of `-clm`, the input is read in chunks and the
+`bytecount` crate is used to count newlines (`-l`) and UTF-8 characters (`-m`).
+In UTF-8 locales, character counting validates the input with `simdutf8` when SIMD
+is enabled, skips malformed sequences, and carries partial characters between
+reads. ASCII chunks are counted by length.
 
 It's useful to vary the line length in the input. GNU wc seems particularly
 bad at short lines.

src/uu/wc/Cargo.toml

@@ -20,6 +20,7 @@
 bytecount = { workspace = true, features = ["runtime-dispatch-simd"] }
 clap = { workspace = true }
 fluent = { workspace = true }
+simdutf8 = { workspace = true }
 thiserror = { workspace = true }
 unicode-width = { workspace = true }
 uucore = { workspace = true, features = [