funny thing, my own v4 has a similar response of low power on 40m CW. interesting you are getting it on a v5. I am getting around 5w or so using a WM2 wattmeter on 40m, maybe just around 10w or so on 80m, and a boost in power as you see on 30 and 20 meters.
I think the v5 retains the toroids feeding the 45 MHz xtal filter as in the v4? now what might these have to do with CW transmit? well I suspect some strange leakage involving these coils and other parts, as the ubitx seeks to 'leak' CW from the PLL direct through the mixer - without mixing as the diodes are biased. as I tried to install a second xtal filter, the 40m leakage matter had more impact and my 40m output dipped a couple watts only.
I am thinking you should enjoy what your nice rig does. If you have an audio oscillator, you might find that you have better SSB output on 40m? I don't see it is worth a lot of work to get from say 5 to 10 watts CW on 40m - its a tiny fraction of an S-unit.
some have replaced those 45 MHz toroids with little (but high enough Q) shielded inductors, curing some spurs that mar the v4 on SSB and as part of the harmonic cure. I have some, but not yet installed (I am enjoying a clean signal on lower bands already).
I tend to see a nice enough v5 you have. it would be good to hear if any other v5's are doing this - but I don't see a huge reason to fix what you have.