pylint failures in travis

[grondo]

On pushes of new branches to travis-ci, I'm seeing failures from pylint that don't really make sense.
@garlick noticed that he got these same failures on his personal branch in travis-ci, but the checks on flux-core/master worked.

The difference here could be that on push to a new branch, all new caches are populated, whereas (I think) the cache for flux-framewor/flux-core/master is reused until manually cleared.

I have no idea, though, what (if anything) from pylint would be cached in travis, so the cache could be red herring. Could anyone with python knowlege help us out?

@trws, @SteVwonder, @morrone?

if [ -x "$( which pylint )" ] ; then  pylint --rcfile=../../../../../../src/bindings/python/.pylintrc ../flux ; fi
************* Module flux.jsc
R: 46, 4: Unnecessary "else" after "return" (no-else-return)
R: 76, 4: Unnecessary "else" after "return" (no-else-return)
************* Module flux.kvs
R: 21, 4: Unnecessary "else" after "return" (no-else-return)
R: 82, 4: Unnecessary "else" after "return" (no-else-return)
R: 90, 8: Unnecessary "else" after "return" (no-else-return)
E:103,12: Bad first argument 'self.__class__' given to super() (bad-super-call)
W:160, 8: __init__ method from base class 'Iterator' is not called (super-init-not-called)
************* Module flux.kz
E: 95,12: Bad first argument 'self.__class__' given to super() (bad-super-call)
************* Module flux.message
E: 25,12: Bad first argument 'self.__class__' given to super() (bad-super-call)
R: 76, 8: Unnecessary "else" after "return" (no-else-return)
************* Module flux.rpc
E: 21,12: Bad first argument 'self.__class__' given to super() (bad-super-call)
************* Module flux.sec
E:  8, 8: Bad first argument 'self.__class__' given to super() (bad-super-call)
************* Module flux.core.handle
E: 26, 8: Bad first argument 'self.__class__' given to super() (bad-super-call)
R: 71, 8: Unnecessary "else" after "return" (no-else-return)

[grondo]

This seems to be caused by a newer version of pylint pulled in by pip on more recent branches. I guess I'll lock down the version of pylint to a known working version (e.g. 1.5.6 instead of 1.7), which will fix this issue, but there is still a problem looming for more recent pylint.