Yes, of course I used a computer. The large numbers involved pose all kind of interesting optimization questions, both w.r.t. time and space. I now have a generator for the list of partitions of n which runs in O

space

. Your algorithm gives:
Or(100) = 2199516347439237513754580009632686701940043714513581388093
(okay, that one still took 5 hours of computer time).
I've seen the sequence at the OEIS. It seems to be the same - at least for the first 30 numbers of that sequence that are listed. Have you tried proving they are the same? Have you submitted your formula to the OEIS?
Technically, your definition of D:
D(alpha) = Prod (i=i..n) ...
still leads to an infinite recursion. If you restrict that one to
D(alpha) = Prod(i=1..n-1)
it doesn't matter if the sum in the definition of Or includes the partition n=n or not.
But that's a technical argument and not a conceptual one. There should be a rationale behind the formulae, and why the partition n=n is included or not. I'll have a look later at your other posts.