MAINT: flake8 *.pyx files

[gfyoung]

closes #12995

flake8-ed *.pyx files and fixed errors.

Removed the E226 check because that inhibits pointers (e.g. char*). In addition, the check is not even universally accepted in Python.

[codecov-io]

Current coverage is 85.26% (diff: 100%)

No coverage report found for master at 3110a72.

Powered by Codecov. Last update 3110a72...5677916

[jreback]

not sure what u mean E226 is not universally accepted

pls turn it back on

[jreback]

ok, forget about E226, its not so easy to remove.

can you add E128 back to the .pxi.in as well.

[jreback]

change this to not use a loop, IOW flake8 pandas --filename '*.pyx' --select=..... will catch all .pyx (even if we add/move them)

[gfyoung]

@jreback : I'm actually surprised that all of these flake8 tests passed. 😄 In any case, do you think we can postpone the pxi.in changes until a separate PR? This pyx change is massive enough as it is.

[jreback]

no this needs fixing here as u edited the .pyx for algos and join which are auto generated and will revert when the pxi.in are changed

[gfyoung]

@jreback : Ah, fair enough. Will fix then.

[gfyoung]

@jreback : On second look, what do you mean that algos.pyx is auto-generated? I believe you mean algos*.pxi are auto-generated. Those are not the same files.

[jreback]

right the .pxi are included at cythonization time
so if u only edited .pyx then ok

[gfyoung]

@jreback : Nice! You got Appveyor to work and now block this PR 😄 😠

At least Appveyor is not failing because of my PR. Need to fix that 3.5 failure with the locale.

[jreback]

oh that's great!

and I stole from numpy a line in appveyor.yml
it cancels the previous if u push again :)
unlike Travis

[jreback]

thanks!