Explain that write! is avoided for perf

Author: teor2345

src/v0.rs

@@ -1,6 +1,7 @@
 use core::convert::TryFrom;
 use core::{char, fmt, iter, mem, str};
 
+// As of 2026, our custom formatting code is significantly faster than using `write!`.
 #[allow(unused_macros)]
 macro_rules! write {
     ($($ignored:tt)*) => {
@@ -412,6 +413,7 @@ impl Disambiguator {
     fn fmt_hex(&self, out: &mut fmt::Formatter<'_>) -> fmt::Result {
         // Handle the edge case of a zero disambiguator, where the bit width is zero, but the
         // format is "0" (not the empty string).
+        // This code is about 4% faster than `write!(out, "{:016x}", self.0)`.
         let hex_len = self.0.bit_width().div_ceil(4).max(1);
         let fill_len = 16 - (hex_len as usize);
         out.write_str(&Self::MAX_HEX_FILL[0..fill_len])?;