Actually, the tooltips are extremely accurate in terms of damage values (oddly). It's pretty safe to use them for such things.
Accounting for accuracy, crit chance, buffs and all procs, I arrived at a vastly different value for the damage-per-force of Double Strike/Thrash:
(1396.0632 + 0.4931(2347.2006 + 365.4873 + 0.33(5241.9175))) / (23 + 0.4931(39 + 10)) = 76.0502
That's actually pretty good efficiency. By comparison, if we look at Project/Shock by itself:
(1700.5211 + 365.4873 + 0.33(5241.9175)) / (39 + 10) = 77.4661
So, slightly lower, but not by much. However, examining abilities in isolation like this is somewhat incomplete and deceptive. The vast majority of threat and damage in a shadow/assassin rotation comes from TkT on proc. Any serious analysis of various abilities needs to carefully consider the effect of adjusting the timing and delay on this ability.
I'd really rather use SimCraft, but as you said, it lacks support for Darkness/Kinetic. I might just roll a one-off simulation script on my own. The abilities and procs are not difficult to encode.
A couple issues with your analysis.
First off, KW/DW cost *10* Force, not 20, per application. It's not a major issue, but it's something you may want to fix in your numbers, especially since your analysis of the hybrid depends upon KW impacting the regeneration.
That is the cost I was using in my analysis. The 20 you see in the numbers is the number of seconds before the charges on KW/DW expire on their own.
You may want to consider trying out your analysis using a hybrid low-Thrash that, rather than try to get the proc routinely (which I only attempt to do when I'm planning on using FP), you simply treat the PA proc as a lucky happenstance that occurs whenever replacing routine SSs with DSs to prevent the waste of Force per cycle pair. Essentially, replace 2 of the 5 Saber Strikes with Double Strikes (1 per cycle so that the chance for a "wasted" PA proc is minimized). The damage and threat will go up without impacting survivability at all (since you're not changing the rate at which HS stacks are generated or TkT is used), which may make it a better "hybrid" than the one you elected to design.
Interesting. I hadn't considered taking the proc but not altering timing based on its presence. I'll redo the analysis on low-thrash with this in mind. I also think that FB/Dsg needs to be considered in analysis like this, but its cooldown is wrong for convenient analysis.