check-accessibility.sh: fix cache bugs

1. Unescaped regex in grep — the url_path was used directly as a grep
   pattern (grep "^$url_path "). The 4 tutorial pages named [1.01]
   Fundamentals..., [1.02] Introduction..., etc. have [ and ] in their
   paths. Grep treats [1.01] as a character class matching 1, ., or 0,
   so the pattern never matches the literal [. This caused:
   - cache_hit to always return miss for those 4 pages
   - cache_update's grep -v to never remove the old entry → duplicate
     entries accumulated (hence 1666 entries for 1616 files)

2. cut -d' ' -f2 fails for paths with spaces — those same tutorial pages
   have spaces in their names. When reading back the hash, cut -d' ' -f2
   returned Fundamentals instead of the hash.

Fix: replaced grep + cut with awk using index($0, p)
(literal prefix match) and $NF (last field = hash).
Also cleaned up the duplicate entries in the cache.

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: ctrueden

_bin/check-accessibility.sh

@@ -54,7 +54,7 @@ cache_hit() {
   [ -f "$CACHE_FILE" ] || return 1
   [ -f "$html_file" ]  || return 1
   current_hash=$(hash_file "$html_file")
-  stored=$(grep "^$url_path " "$CACHE_FILE" 2>/dev/null | cut -d' ' -f2)
+  stored=$(awk -v p="$url_path " 'index($0,p)==1{print $NF;exit}' "$CACHE_FILE" 2>/dev/null)
   [ "$stored" = "$current_hash" ]
 }
 
@@ -65,7 +65,7 @@ cache_update() {
   [ -f "$html_file" ] || return
   current_hash=$(hash_file "$html_file")
   tmpfile=$(mktemp)
-  grep -v "^$url_path " "$CACHE_FILE" 2>/dev/null > "$tmpfile" || true
+  awk -v p="$url_path " 'index($0,p)!=1' "$CACHE_FILE" 2>/dev/null > "$tmpfile" || true
   echo "$url_path $current_hash" >> "$tmpfile"
   mv "$tmpfile" "$CACHE_FILE"
 }