PyCBC live results should record when followup p-value was used and trials factors

[tdent]

For interpreting/improving the multi-ifo behaviour it is useful to know - ie record - whether the with-followup IFAR was larger and in any case to record the trials factor and/or pre-trials IFAR. Currently the p-value combination function doesn't leave any record of which IFAR was higher, it just takes the max and divides by 2.