pr: uniformly scan for form feed and newline chars

Fix the way form feed characters are interpreted by changing the way
lines and pages are found. Before this commit, a file comprising two
form feed characters (`/f/f`) would result in too few trailing newlines
at the end of the second page. After this change, each page is produced
with the correct number of lines.

This commit changes the way files are read, replacing complex iterators
with a loop-based approach, iteratively scanning for newline or form
feed characters. The `memchr` library is used to efficiently scan for
these two characters. One downside of this implementation is that it
currently reads the entire input file into memory; this can be improved
in subsequent merge requests.

Cargo.lock

@@ -3690,6 +3690,7 @@ dependencies = [
  "clap",
  "fluent",
  "itertools 0.14.0",
+ "memchr",
  "regex",
  "thiserror 2.0.17",
  "uucore",

src/uu/pr/Cargo.toml

@@ -21,6 +21,7 @@ path = "src/pr.rs"
 clap = { workspace = true }
 uucore = { workspace = true, features = ["entries", "time"] }
 itertools = { workspace = true }
+memchr = { workspace = true }
 regex = { workspace = true }
 thiserror = { workspace = true }
 fluent = { workspace = true }

tests/by-util/test_pr.rs

@@ -606,3 +606,28 @@ fn test_omit_pagination_option() {
         .pipe_in("a\nb\n")
         .succeeds();
 }
+
+#[test]
+fn test_form_feed_newlines() {
+    // Here we define the expected output.
+    //
+    // Each page should have the same number of blank lines before the
+    // form-feed character.
+    let whitespace = " ".repeat(50);
+    let datetime_pattern = r"\d\d\d\d-\d\d-\d\d \d\d:\d\d";
+    let page1 = format!("\n\n{datetime_pattern}{whitespace}Page 1\n\n\n\n\x0c");
+    let page2 = format!("\n\n{datetime_pattern}{whitespace}Page 2\n\n\n\n\x0c");
+    let pattern = format!("{page1}{page2}");
+    let regex = Regex::new(&pattern).unwrap();
+
+    // Command line: `printf "\f\f" | pr -f`.
+    //
+    // Escape code `\x0c` in a Rust string literal is the ASCII escape
+    // code `\f` for the "form feed" character (which appears like
+    // `^L` in the terminal).
+    new_ucmd!()
+        .arg("-f")
+        .pipe_in("\x0c\x0c")
+        .succeeds()
+        .stdout_matches(&regex);
+}