solver tries to call endRevealPeriod twice

[hswick]

I noticed that on the second run of a test that SOLVER is trying to call end reveal period, this occasionally leads to an error. The error doesn't That doesn't affect the test outcome, but it is still concerning. It may be the weird revert we've seen on kovan as well

Can be replicated by running npm run test test/os-wasm-challenge.js a few times back to back