Handle CW prosigns correctly in renderer/morse.py

Prosigns get split into individual letters with an (incorrect) gap
between them. E.g. 'SK' is rendered as '... -.-' but shoud be '...-.-'.
This also preserves `[morse:correction]` by producing a series
of nine dits.

The patch proposed by HB9ARK fixes that. Thank you!

renderer/morse.py

@@ -78,11 +78,36 @@ class Morse(SpanToken):
             " ": [space],
         }
 
+        # Prosigns (e.g. "ar", "bk", "sk", "correction") are stored as multi-
+        # character keys whose dit/dah list already has no gap between the
+        # individual letters - that's what makes them a prosign rather than
+        # two separate letters. They must therefore be matched as a whole
+        # "word", not character by character, otherwise they're never found
+        # and get split into individual letters with an (incorrect) gap
+        # between them.
+        prosigns = {key: value for key, value in morse_code.items() if len(key) > 1}
+
         result = []
 
-        for char in text:
-            if char.lower() in morse_code:
-                result.append(morse_code[char.lower()])
+        # Split into words and whitespace, keeping the whitespace as its own
+        # token so word gaps are preserved exactly as before.
+        for token in re.split(r"(\s)", text):
+            if not token:
+                continue
+
+            if token.isspace():
+                for char in token:
+                    if char in morse_code:
+                        result.append(morse_code[char])
+                continue
+
+            if token.lower() in prosigns:
+                result.append(prosigns[token.lower()])
+                continue
+
+            for char in token:
+                if char.lower() in morse_code:
+                    result.append(morse_code[char.lower()])
 
         return result