Fix regression with ascii_identifiers and Turkish undotted i.

Closes #8003.

src/Text/Pandoc/Asciify.hs

@@ -18,11 +18,15 @@ import Data.Text (Text)
 import qualified Data.Text as T
 
 toAsciiText :: Text -> Text
-toAsciiText = T.filter isAscii . TN.normalize (TN.NFD)
+toAsciiText = T.filter isAscii . T.map specialCase . TN.normalize (TN.NFD)
+ where
+  specialCase '\x131' = 'i' -- Turkish undotted i
+  specialCase c = c
 
 toAsciiChar :: Char -> Maybe Char
 toAsciiChar c = case T.unpack (TN.normalize TN.NFD (T.singleton c)) of
                   (x:xs) | isAscii x
                          , all isMark xs
                          -> Just x
+                  ['\x131'] -> Just 'i'  -- Turkish undotted i
                   _      -> Nothing

test/command/8003.md

@@ -0,0 +1,6 @@
+```
+% pandoc -f markdown+ascii_identifiers
+# Işık
+^D
+<h1 id="isik">Işık</h1>
+```