Provide a sane result for manual and level depth setting

Specifying level depth while using manual mode does not have the context
to provide a relative number and is not recommended. But if it is used,
just return a value of all ones at the specified depth, e.g. 1, 1.1,
1.1.1, etc.

pymdownx/blocks/caption.py

@@ -288,6 +288,9 @@ def on_end(self, block):
 
         prefix = self.PREFIX
         if prefix and not self.auto:
+            # Levels should not be used in manual mode, but if they are, give a generic result.
+            if self.level:
+                self.fig_num = '.'.join(['1'] * int(self.level))
             if self.fig_num:
                 update_tag(
                     block,