That's very good news about tap-tempo development.
I found this thread after I mentioned elsewhere http://www.diystompboxes.com/smfforum/index.php?topic=79078.msg666768
(Gristleizer PCB's) that a soft LFO might be worth pursuing, and then I found that it had already been done.
Does the tap-tempo input completely replace the frequency control? If it does, then you can't use the pot to nudge the frequency up or down, you have to re-tap it. And if it doesn't and there's still a frequency pot too, then there is the usual problem of editing an existing value when the pot's physical position doesn't match the setting. Either you do a Prophet-5 style edit from where you happen to be, or you use a rotary encoder and accept the finite steps which are larger than the smallest step on a pot.
I was thinking about the algorithm. Taps 'expire' after 65 seconds, which is an absolute age in messing around with things terms. Consider the situation where it is 64 seconds since the last tap. You tap twice. The first tap is paired with the ancient tap, setting the frequency to very-slow-indeed. The second tap is now taken as a new 'first' tap, so noted but not acted on. The net result is that the frequency is still very-slow-indeed. Now consider the situation where it is 66 seconds since the last (unpaired) tap. The first tap is noted, the second concludes the timing interval and the LFO is set to the tempo defined by those two new taps. That is very different behaviour, depending on stuff that happened ages ago.
Would it be better to not differentiate between 'first' and 'second' taps, and each time there is a tap, try to pair it with a preceeding tap and set frequency appropriately. If there is no preceeding tap within the last 65 seconds, note the new tap but do nothing to the frequency yet.
If you're really clean about never leaving an unpaired tap it would never be a problem. I know I'm not like that. I might tap it three times, five times, whatever, until it sounded right.
I can't pretend to speak from first hand experience, though. This is just conjecture. How does it pan out in practice?