A brief update on the SRCL dither arm power measurement results:
After the EX to IX ring heater TCS move, I ran another SRCL dither arm power test.

DC measurements vs plane-wave model
I've estimated the arm powers via the carrier reflection (alog 66432)
the estimated CARM pole, and the PRG (alog 65095).

Each of these reported that the arm power was somewhere between 300 and 337 kW, with round trip losses between 58 and 69 ppm.  
This assumes a plane-wave, perfectly mode-matched model.

The uncertainties on each of these types of loss measurements is probably on the order of 5-10%, and this is strong, varying dependence of the arm power on round trip loss. 
The fact that these three measurements hang together is fairly compelling, and if I were to do a rigorous analysis would probably constrain our arm power measurement to around 320 kW +- 10 kW or so, (with around 47 W input power on PRM, and measured PRG of around 54).

Radiation pressure based measurement
However, since May 2022, the SRCL dither arm power measurement has not hung together with any of these measurements.
The principle behind the SRCL dither measurement is we modulate the power in each arm differentially by moving the SRM, which modulates the DARM offset light in the SRC.
This modulated DARM offset light returns to the arms, one arm receiving a sign flip from the beamsplitter.
This creates a strong differential radiation-pressure based DARM modulation, which is easily measurable.
When compared to the transmitted power relative intensity noise, the DARM / ARM RIN TF should give you a direct measurement of arm power (DCC T1900529).
(The SRCL to DARM radiation-pressure coupling was known in the design of LIGO and is why SRCL motion is critical to suppress).

In May 2022, I found an arm power of around 243 kW +- 1 kW with some very clean data.
In some sense, this is very hard to refute, because the data is clean, the light in the arms is clean, the measurement is simple and probes the TEM00 light in the arms directly, and the model simple in that is only relies on the mass of the optics and the power in the arms.
You need to be sure you are in the radiation-pressure dominated regime as well, but that is simple enough by finding the f^-2 coupling region for DARM/ARM RIN.

I did the SRCL dither measurement again on January 16, 2023. (60 W input, 51.5 PRG)
My first injections were not very strong, and I was not sure that in the nominal state we run in that the DARM/ARM RIN coupling was following f^-2.
I increased the DARM offset such that the DCPD light was 40 mA, instead of the nominal 20 mA, to increase our SNR.
I found an arm power of around 
X Arm Power Estimate = 268.3 +- 1.8 [kW]
Y Arm Power Estimate = 261.2 +- 1.9 [kW]

It's possible that DARM was not well-calibrated because I changed the DARM offset, but to first order the DARM loop math which keeps the UGF the same should take into account any DARM offset changes.
I've checked and re-checked the arm QPDs, they don't seem to be saturated or anything.  The TMS SERVOS are always on now, and the SRCL and MICH feedforward was off.

Results
As things stand, these results still do not hang together.
I've discussed this with Dan Brown some, and he may have some simulations about mode-matching and the thermal state in CARM creating conditions where this direct radiation-pressure probe of the arm power is consistent with the DC power estimates.
As usual, these models quickly become so complicated it's very hard to verify if they represent anything real.
More modeling is needed.