lib: banner purge + WHY-only comment sweep (rule 8)

Every // ---- section banner dies (132 -> 0): names carry the WHAT.
Narration deleted (filename restatements, run-with lines, what-the-
next-line-does); every ref/hare cite, task cite, divergence, ABI/
layout contract, and ownership qualifier kept (borrowed-view lines
restored where the sweep over-cut). Comment-only proven: all 442
walk-workdir .s and 32 import-probe .s byte-identical before/after;
libbyteid 56-roster all-ID.

lib/hash/adler32/adler32.ww

@@ -1,4 +1,4 @@
-// hash/adler32 — Adler-32 checksum (RFC 1950). Pure ww.
+// Adler-32 checksum (RFC 1950).
 //
 // Hare ships a hash::hash-shaped streaming type backed by io::stream;
 // we ship the pure-buffer subset here, same shape as lib/hash/fnv.

lib/hash/crc16/crc16.ww

@@ -1,5 +1,3 @@
-// hash/crc16 — CRC-16 checksum. Pure ww.
-//
 // Inline polynomial-shift per byte (no precomputed tables). Slower
 // than a table-driven CRC by ~8x per byte but matches the
 // table-driven answer bit-for-bit.
@@ -15,9 +13,7 @@ def ANSI: u16     = 0xA001u16;	// Modbus, USB, ANSI X3.28
 
 // sum16 — fold `buf` under `poly` and return ~cval. Initial value is
 // ~0u16, matching the streaming CRC-16 contract for a single
-// write-then-sum. Per byte: XOR low byte of cval with msg byte to form
-// an 8-bit index, run 8 polynomial shifts on that index, XOR the
-// result with the high byte of cval shifted down.
+// write-then-sum.
 export fn sum16(buf: []u8, poly: u16) u16 = {
 	let c: u16 = 0xFFFFu16;
 	let i: i32 = 0;

lib/hash/crc32/crc32.ww

@@ -1,5 +1,3 @@
-// hash/crc32 — CRC-32 checksum. Pure ww.
-//
 // Same shape as lib/hash/crc16: per-byte inline polynomial shift,
 // no precomputed table. Slower than Hare's table-driven path by ~8x
 // per byte but produces identical answers.
@@ -10,9 +8,7 @@ def IEEE: u32       = 0xEDB88320u32;	// gzip, PNG, zip, Ethernet
 def CASTAGNOLI: u32 = 0x82F63B78u32;	// iSCSI, SCTP, SSE4.2
 def KOOPMAN: u32    = 0xEB31D82Eu32;	// small datasets
 
-// sum32 — fold `buf` under `poly` (reversed form). Initial cval is
-// ~0u32; per byte we mix in the low byte via 8 polynomial shifts and
-// XOR with the high three bytes shifted down.
+// sum32 — fold `buf` under `poly` (reversed form). Initial cval is ~0u32.
 export fn sum32(buf: []u8, poly: u32) u32 = {
 	let c: u32 = 0xFFFFFFFFu32;
 	let i: i32 = 0;

lib/hash/crc64/crc64.ww

@@ -1,5 +1,3 @@
-// hash/crc64 — CRC-64 checksum. Pure ww.
-//
 // Same shape as lib/hash/crc32: per-byte inline polynomial shift, no
 // precomputed table. Slower than Hare's table-driven path by ~8x per
 // byte but produces identical answers for the documented polynomials.
@@ -11,9 +9,7 @@ package crc64;
 def ECMA: u64 = 0xC96C5795D7870F42u64;	// ECMA-182, xz-utils
 def ISO:  u64 = 0xD800000000000000u64;	// ISO 3309 HDLC
 
-// sum64 — fold `buf` under `poly` (reversed form). Initial cval is
-// ~0u64; per byte we mix in the low byte via 8 polynomial shifts and
-// XOR with the high seven bytes shifted down.
+// sum64 — fold `buf` under `poly` (reversed form). Initial cval is ~0u64.
 export fn sum64(buf: []u8, poly: u64) u64 = {
 	let c: u64 = 0xFFFFFFFFFFFFFFFFu64;
 	let i: i32 = 0;

lib/hash/fnv/fnv.ww

@@ -1,5 +1,3 @@
-// hash/fnv — FNV-1a 64-bit. Pure ww. No dependencies.
-
 package fnv;
 
 def OFFSET: u64 = 14695981039346656037;

lib/hash/siphash/siphash.ww

@@ -1,5 +1,3 @@
-// hash/siphash — SipHash-2-4 keyed hash, buffer-based.
-//
 // Mirrors Hare's hash::siphash for the one-shot path: take a 16-byte
 // key and a buffer, return the 64-bit hash. Hare ships a streaming
 // io::stream-backed type; ww's subset doesn't, matching the rest of