Attempt at better guess_lexer_for_filename

the idea: guess on filename first, and on code second.
prompted by: getting a Django Template Parser to work.

however, this won't fly, because lexer.analyse_text is basically useless
(the parts that are implemented __at all__ are often based on some initial
bytes in the file, which is useless in our context of code snippets)

theme/templatetags/issues.py

@@ -35,7 +35,7 @@ def _core_pygments(code, filename=None, platform=None):
     # code is.
 
     if filename:
-        lexer = guess_lexer_for_filename(filename)
+        lexer = guess_lexer_for_filename(filename, code=code)
         if lexer is None:
             lexer = lexer_for_platform(platform)
     else: