Remove quadratic time complexity in Statement.cast()

Using sum(lst, []) to flatten a list of lists has quadratic time
complexity. Use chain.from_iterable() instead. While not strictly
necessary to improve performance, convert to map().

A test case writing out verilog for a 512k entry FIFO is 120x faster
with this applied.

nmigen/hdl/ast.py

@@ -4,6 +4,7 @@
 from collections import OrderedDict
 from collections.abc import Iterable, MutableMapping, MutableSet, MutableSequence
 from enum import Enum
+from itertools import chain
 
 from .. import tracer
 from .._utils import *
@@ -1404,7 +1405,7 @@ def __init__(self, *, src_loc_at=0):
     @staticmethod
     def cast(obj):
         if isinstance(obj, Iterable):
-            return _StatementList(sum((Statement.cast(e) for e in obj), []))
+            return _StatementList(list(chain.from_iterable(map(Statement.cast, obj))))
         else:
             if isinstance(obj, Statement):
                 return _StatementList([obj])