Hmmm...
Cross: B' L2 U R' F2 (5)
1st: L U L' U B' U' B (7)
2nd: F' U F U2 F U F' (7)
3rd: F' U' F U' R' U' R2 (7)
the R2 instead of R in anticipation of...
4th: U2' R' U' R U R' (6)
OLL: U' R U R' U R d' R U' R' F' (11)
PLL: U' x U' R U' L2 U R' U' L2 U2 (11)
Total: 54 moves
For that 1st insertion case that you did I use d' R U R' U2' R U R' U' R U R', it's 11 moves, but it's hella fast when you push the U2' around with your right index. It's one of the cases I don't do intuitively. I've just picked up F2L algorithms here and there as I find I don't like the solution I have, I didn't sit down and grind them like my OLLs, but I probably have about a dozen or more in my repertoire now.
If I'd picked your pair in inserted it my way I would have continued with:
2rd: U2 R' U R U' B U' B'
3rd: U' L' U L2 U' L'
4th: L' U L U' F U F'
OLL: F U R U' R' F'
PLL: U' Skip
Which would have rocked.
Ok already 11 is way too many moves to insert an F2L pair especially with a cube rotation in the middle. For nasty cases like that you should learn an alg.
You've been doing OH for a long time Lofty. ;-)
For edge correct, corner on top of it with D color up, I do (R U R' U') (R U R' U') (R U R'), which even my old, slow bones can do in about 1.6-1.8 seconds. But, yes, an intuitive insert that takes 11 turns *and* a cube rotation, definitely needs an alg. The alg I mentioned above takes me about 2.2 seconds, which is really appalling, but I'm pretty slow.