Fix #139: newlines in async for comprehension

[Kazy]

I decided to take a stab at this. I don't know if this is the right fix, and if the test is the right place (should it be in test_tokenize.py ?).

I traced back the difference between the parsing of a for comprehension and an async one to the tokenization. In a normal for-comprehension, the ] token has \n as a prefix, while in the case of async for it has NEWLINE token which isn't part of the DFA transitions and is thus a SyntaxError.

Removing async from the ALWAYS_BREAK_TOKENS fixes this, all the tests are passing. I don't know if this can have consequences in other uses of async. Codes like this are still rejected:

async
def foo():
   ....

# or
async
    def foo():
        ...

Now it also correctly parses all possible insertion of newlines in the async for-comprehension, just like a sync one.

async def foo():
    [1 
    async
    for
    _
    in
    range(5)
    ]

[isidentical]

LGTM and Thank you for your contribution. (for a better look on the test, if you want you can move the last triple quotes to the upper line (so it wont break the indentation)).

[Kazy]

(for a better look on the test, if you want you can move the last triple quotes to the upper line (so it wont break the indentation))

Done !