I'm on the fence about the CA-2A upgrade. 
I have the LA-2A and like it, but for me it seems the only benefit of the upgrade is side chaining.
Now I know you can use it in the fx bin but that really doesn't drive the decision for me.
I am fine with the PC use. 
For me it really comes down to the issue of using side chaining.
I have never used it.
Maybe just my lack of experience showing here.
I understand the concept.

Am I missing out on something really huge by not doing side chaining?

Appreciate any insights.

Dave

Funny thing is, in the old days of hardware and large confusing patchbays I used side-chains a LOT, heck, I modified gear to provide side-chains and key inputs. They were an important part of the arsenal! One of my favorite tricks was placing a reverb or delay in the sidechain or key input for a dynamics processor<G>!

With DAWs there are so many ways to control processors that I don't find myself reaching for side-chain patch cables as often. More often I'll set up automation so that I can control the behavior of the processor in a more deterministic way.

But I do still use side-chains and key inputs, and I'm looking at the CA-2A for that very reason.

If you're just looking for sidechaining capability, you already have it: the Sonitus compressor will serve most sidechain applications just fine. Given that you already have an LA-2A emulation, and that you can already do sidechained compression, the CA-2A would seem to be redundant. Personally, I'd spend that money on something that filled some other, currently unfilled, niche in my toolkit. 

As far as how useful sidechaining is, if you're doing EDM or hip-hop or some other beat-oriented genre, then it's intrinsic to the style and employed routinely. For other genres it's a nifty special effect but far from necessary or routine. Either way, it's a technique you should become familiar with. I'd suggest having some fun experimenting with sidechaining the Sonitus compressor, and then decide if you need to specifically sidechain an opto-style compressor.