Issue #24619: New approach for tokenizing async/await.

This commit fixes how one-line async-defs and defs are tracked
by tokenizer.  It allows to correctly parse invalid code such
as:

>>> async def f():
...     def g(): pass
...     async = 10

and valid code such as:

>>> async def f():
...     async def g(): pass
...     await z

As a consequence, is is now possible to have one-line
'async def foo(): await ..' functions:

>>> async def foo(): return await bar()
diff --git a/Lib/tokenize.py b/Lib/tokenize.py
index 3ec9018..c3efdda 100644
--- a/Lib/tokenize.py
+++ b/Lib/tokenize.py
@@ -501,6 +501,7 @@
     # 'stashed' and 'ctx' are used for async/await parsing
     stashed = None
     ctx = [('sync', 0)]
+    in_async = 0
 
     if encoding is not None:
         if encoding == "utf-8-sig":
@@ -580,6 +581,9 @@
 
                 cur_indent = indents[-1]
                 while len(ctx) > 1 and ctx[-1][1] >= cur_indent:
+                    if ctx[-1][0] == 'async':
+                        in_async -= 1
+                        assert in_async >= 0
                     ctx.pop()
 
                 yield TokenInfo(DEDENT, '', (lnum, pos), (lnum, pos), line)
@@ -640,7 +644,7 @@
                         yield TokenInfo(STRING, token, spos, epos, line)
                 elif initial.isidentifier():               # ordinary name
                     if token in ('async', 'await'):
-                        if ctx[-1][0] == 'async' and ctx[-1][1] < indents[-1]:
+                        if in_async:
                             yield TokenInfo(
                                 ASYNC if token == 'async' else AWAIT,
                                 token, spos, epos, line)
@@ -657,6 +661,7 @@
                                 and stashed.string == 'async'):
 
                             ctx.append(('async', indents[-1]))
+                            in_async += 1
 
                             yield TokenInfo(ASYNC, stashed.string,
                                             stashed.start, stashed.end,