I fixed several bugs in birdie generation. One of these involved the time spacing of signals: this needs to be at least the FFT duration. Another was yet another confusion of time units (days and seconds... I wish we had used seconds throughout!).

After fixing these bugs, the number of birdie signals decreased by about 10X. Birdies no longer dominate the top 100 multiplets - in fact there's only one birdie multiplet in the top 10,000. This doesn't seem right, and we'll investigate it.

In the meantime, we can investigate the top multiplets, which are now RFI, ET, or noise rather than birdies. I'll write about this separately.

-- David