Parameter for the epoch window width of cubic slashing

[brentstone]

We currently have in the Specs:

This issue tracks whether or not we want to do this with a parameter and if it would belong in the protocol or PoS params.

[cwgoes]

It should be symmetric, otherwise there are weird effects based on when slashes are detected. (-n, +n) would be reasonable, with n the parameter.

[brentstone]

Completed in #892