Fix a bug causing both W503 and W504 to be ignored

[novadev94]

This happens when either W503 or W504 is in the ignore list, autopep8 decided to ignore both of them - which it shouldn't.

[coveralls]

Coverage increased (+0.005%) to 98.639% when pulling 7984fe7 on NovaDev94:master into fded37b on hhatto:master.

[hhatto]

Thanks for contribution 👍

What problem does this change solve?
Could you show us specific Python code for autopep8 and the execution image of autopep8?

Also, could you add some test code for this change?
That will help us better understand the intent of the change.

---

related issue: #456, #466

[novadev94]

@hhatto Sorry I forgot to include an example

# setup.cfg
[flake8]
ignore = W503
# So autopep8 should format based on W504 (break before paramemter)

# test.py
a_bool = (
    this and
    that
)

With the current autopep8 behavior, it doesn't re-format it into

# test_formatted.py
a_bool = (
    this
    and that
)

Same behavior (doesn't reformat the operator position) can be seen when we ignore W504 (but not W503).

This is caused by the not all(any(conflict in ignores) for conflict in (W503, W504)) forces both codes to be ignored even when one of them already was.

| W503 | W504 | Ignore | Correct |
|:----:|:----:|:------:|:-------:|
|      |      |  Both  |    Y    |
|   x  |      |  Both  |    N    |
|      |   x  |  Both  |    N    |
|   x  |   x  |  Both  |    Y    |

(The last case is not needed since both of them were already ignored) 

I'm using all(not any(...)) to only add those 2 codes to the ignore list when none of them already was.

[novadev94]

Unit tests added

[novadev94]

@hhatto I've squashed the unit test commit into the main one. Please let me know if you need extra information about the scope of this change.

[hhatto]

LGTM 👍

Thank you for contribution!!