This feat is notorious for its poor wording. The “+100%” phrasing is completely unique within D&D 3.5e as far as I know, for example. Ultimately, I can’t imagine any other interpretation here than adding again the number subtracted from your attack rolls, and it does have the nice feature of specifying the “normal” damage from Power Attack which means that features like the frenzied berserker’s supreme power attack that already give one-handed weapons 2:1 returns don’t get doubled to 4:1, but instead go to the 3:1 you would normally expect from D&D’s multiplication rules.
But then there is the line you haven’t quoted:
If you use this tactic with a two-handed weapon, you instead triple the extra damage from Power Attack.
No bizarre “+100%” in sight! But also we have lost the useful reference to “normal” and now it is multiplying “the extra damage from Power Attack,” whatever that is for you. This is going to get us in trouble, you can just tell already.
So you are tripling the extra damage—not tripling the penalty applied. The problem here, well the first problem here, is that “the extra damage from Power Attack” is “twice the number subtracted from your attack rolls” when attacking two-handed. Worse, since “the extra damage from Power Attack” is calculated as twice the penalty, but isn’t itself subject to any multiplier, arguably the repeated-multiplication rules don’t apply, and that gets you a 2×3=6 rather than 1+(2−1)+(3−1)=4. So instead of 2:1 returns on Power Attack, you get 6:1 returns on Power Attack. Or maybe you get 5:1; it’s impossible to say since it’s worded so poorly. Plus, ya know, I suspect what they meant to do was give you 3:1 returns, but of course they didn’t say that.
And that would combine quite nicely with, say, the supreme power attack feature of the frenzied berserker, who was getting 4:1 returns to begin with. Now they’re arguably getting 8:1.
On top of those issues, this is only the Power Attack bonus damage. The result is added to the rest of your damage, and that gets you your full damage... which might be multiplied again, e.g. with valorous. This effectively multiplies your multiplier, which is exactly what the multiplication rules try to avoid, but since two different things are being multiplied, the multiplication rules don’t actually come into play.
So for the example: 2d6+1 damage from the weapon itself, +6 for Strength, and the −6 attack penalty for maximum Power Attack results in double that for +12 damage from Power Attack without Leap Attack. Thus 2d6+19 is the baseline for all interpretations, and valorous doubles that for 4d6+38.
With the 6:1 returns, we are instead looking at Power Attack bonus of +36 (six times the penalty, triple “the extra damage from Power Attack” which would have been +12). Using 5:1 brings that down to +30, which is somewhat better, but not, ya know, great, when what they probably meant was +18. Note that +36 is nearly what valorous was giving the entire attack before. Now with valorous, we’re looking at a total of 4d6+66—of which, 52 comes from Power Attack.
It may not be a bad idea to try to eliminate the multiplication of a multiplier here through houserule, but note that the Power Attack bonus damage isn’t the only case of this: the bonus damage due to Strength also has a multiplier, +1½×, which is also being doubled by valorous. This, unlike Leap Attack, has strong precedent in the rules. The “fix” would be to apply the multiplication rule individually to all sources of damage, like so:
\begin{array}{r}
2 \times ( && 2\text{d}6 && +1 && +1\tfrac{1}{2}\times 4 && +3\times 2\times 6 & ) \\
= && 2\times 2\text{d}6 && + 2\times 1 && + 2\times 1\frac{1}{2}\times 4 && + 2\times 3\times 2\times 6 \\
= & [1 \\
&& +\left(2-1\right) \\
& ] & \times 2\text{d}6 & +[1 \\
&& && +\left(2-1\right) \\
&& & ] & \times 1 & +[1 \\
&& && && +\left(2-1\right) \\
&& && && +\left(1\frac{1}{2}-1\right) \\
&& && & ] & \times 4 & +[1 \\
&& && && && +\left(2-1\right) \\
&& && && && +\left(3-1\right) \\
&& && && && +\left(2-1\right) \\
&& && && & ] & \times 6 \\
= && 2\times 2\text{d}6 && +2\times 1 && +2\frac{1}{2}\times 4 && +5\times 6 \\
= && 4\text{d}6 && +2 && +10 && +30 \\
= && && && && 4\text{d}6+42 \\
\end{array}
But this is very-definitely a houserule, and I’m not convinced that it is good (I mean, good luck calculating that for every attack!), even though it “enforces” the idea that you’re not supposed to get to mulitply multipliers.