gsathya / metrics-tasks Public

Suggested by atagar.

Last modified times of .csv files do not always match publication dates.

File header now contains database dates and file names.

Calculating the maximum probability difference between pristine and empty
consensus doesn't make much sense.  More importantly, the current
implementation violates the function relationship between linf and
min_adv_bw, because we're not increasing min_adv_bw in the last run and
thereby outputting two different linfs for the same min_adv_bw.  Should be
safe to omit that final run, because we're not looking for the case where
we remove all relays from the consensus anyway.  I hope.

Relays shouldn't be excluded based on their position-dependent weighted
consensus weight, but on their consensus weight directly taken from the
consensus.  No multiplying by Wxx weights necessary for that.

The current for loop to remove routers with min cw removed only every
second element.  Use a while loop instead.