Bug: Divide by zero in sparse monthly log trend analysis

If only one weight was present in the most recent log and no entries existed in the previous month's log for the preceding week, the {\tt trendfit::fitSlope()} method would divide by zero because there were insufficient points to fit a linear trend. I added code to the method which reports a zero trend slope when insufficient points are available to fit a slope. Note that while this was discovered testing the new method of computing the trend for the current month, the bug was present prior to the change.