3BLD pure flip/twist stats

There was a discussion recently about the probabilities of flipped edges in scrambles, and I decided to do the calculations. I did it for edges and corners, and for including buffer flips (column called "buffer") and the more relevant figures for ignoring buffer ("true"). I'll list the number of cases of each (adds up to 12!*2^11 and 8!*3^7 as appropriate), and the probabilities (aggregating once it gets below 0.001%). Hopefully I didn't screw up, the results seem sensible enough. I don't know if anyone has done these exact statistics before, if so I can check my numbers I guess, if not someone could maybe check my numbers if they feel like it.

edges
 n       buffer         true
 0 595003712413 619795533763
 1 297501856200 285105945531
 2  74375464086  65078531045
 3  12395910560   9813429285
 4   1549489095   1097554590
 5    154948464     96843054
 6     12912900      6994218
 7       921888       422730
 8        57915        21615
 9         3080          935
10          198           33
11            0            1
12            1            0

 n buffer   true
 0 60.653 63.180
 1 30.327 29.063
 2  7.582  6.634
 3  1.264  1.000
 4  0.158  0.112
 5  0.016  0.010
6+  0.001  0.001

corners
n   buffer     true
0 45273046 49045792
1 30181968 28924448
2 10060904  8383872
3  2235184  1583680
4   373380   217280
5    48944    22848
6     5992     1792
7      336      128
8       86        0

 n buffer   true
 0 51.342 55.620
 1 34.228 32.802
 2 11.410  9.508
 3  2.535  1.796
 4  0.423  0.246
 5  0.056  0.026
6+  0.007  0.002