Since I have done a combinatorial analysis I thought I may as well share the results, even though I am sure the conclusions are intuitive for most of you.
I first did a calculation for Spanish 21 as if it was the first hand in a shoe.
This ignored hands where the dealer had an A in the hole because the option is not available in this instance.
To take this option your EV should be below 0.5
for 10,10 vs 10 your EV is 0.597
for A,9 vs 10 your EV is 0.593
I then did the calculation with 2 decks, to see how much the effect changed during the shoe for the same true count.
for 10,10 vs 10 your EV is 0.609
for A,9 vs 10 your EV is 0.593
So, this option gets even worse for you as the shoe progresses for 10,10 vs 10.
So, I wondered if there would be a count where the option would become a good one. I worked out the EORs for each card type for this situation. They are as follows (rounded to 3 dp).
A: -0.001
2: -0.003
3: -0.004
4: -0.003
5: +0.001
6: +0.001
7: -0.006
8: -0.007
9: -0.009
10: +0.010
As you can see, in respect to using an unblanced hi-lo count, this poses all sorts of problems. First, the 3 cards that have the most effect in reducing the EV in this situation (making it more likely to take the bet) just happen to be the 3 cards that hi-lo counts as 0. Secondly, if you look at the +1 cards (2 through 6) you can see that 3 of them have a - effect and 2 have a + effect, making it rather useless as a way to judge this bet. There were also interesting effects of multiple card removal, for instance removing a 5 and a 6 from each deck, although individually improving the EV, left the EV the same.
There is no real point using a hi-lo count where this option becomes +EV. If we remove all 32 of the 7s, 8s and 9s (the cards with the biggest effect) from an 8 deck shoe your EV becomes 0.491, making the option a 0.009 EV favourite. So, if you use a rediculous side count and wait for an impossible situation, you may get to make 90c in EV on a $100 bet.
I did a similar analysis for regular BJ with similar reaults, although the EV for 10,10 vs 10 off the top of an 8 deck shoe is lower at 0.558 due to the increased chance of pushes with the extra 10s in the deck.
All together, this is all pretty obvious intuitively, but was good practice for my blackjack math so I enjoyed it anyway. If there are any massochists who would like to see the spreadsheet just let me know too.