So, fwiw, some latency numbers:
Dream, Enigmatic, Lion, Showtime amps: 34 samples.
Ruby: 37
Woodrow: 40
RAW, Big Fuzz, Vintage Fuzz: 22
Gold, TS, NVL: 5
Red Comp: 22 (interestingly enough, 0 for the 1176)
Blue Flanger: 11
Phaser, Trem: 5
224 Reverb: 1
Anything else (cabs included, which actually made me wonder a bit - but perhaps that's already accounted for with the amps, who knows...) doesn't add any latency on top.
So, a chain of, say, Compressor, RAW drive, Blue Flanger and Woodrow (not exactly too esoteric) would result in 95 samples of additional latency at 44.1kHz, which is around 2.1ms.
Bypassing any block in your chain will not change the latency it adds, you'll have to remove the block in order to get rid of the additional samples.
And in case someone's interested: The amount of samples scales up with the samplerate set in your DAW. So when you run things in 88.2kHz, there will be the double amount of additional samples, hence the same resulting latency.
---
Are these really bad numbers? It absolutely depends but I'd tend to say yes, especially as others don't suffer from this (and for comparisons sake: My entire floor pedal with 4 full ADDA cycles only kicks in at a tad over 3ms).
Personally, I think they should add a "zero latency mode" that you could switch to (such as, say, Acustica Audio do). That'd cause higher CPU usage, but would allow you to always stay within the latency values determined by the buffer settings you're using in your DAW.