The "maximum theoretical power consumption" depends on device junction temperatures, thermal conductivity, air flow, etc, etc. I don't think that's what your question is really after - but if it is, you can stop reading here.
The amount of power the RPi 4B consumes "under load" is regulated by the device's firmware in a closed-loop system. In other words: At some measured value of device temperatures (CPU, GPU, PMIC), the firmware acts to "throttle" performance to limit further increases in the temperature. We don't know precisely how that is done, because the Raspberry Pi firmware is proprietary. For this reason, empirical limits of power consumption are of more practical interest than theoretical limits.
Test measurements from 4Q 2019 conducted by "The Organization" on the then-current firmware showed that the RPi 4B power dissipation tops out at approximately 6.4 Watts running with the latest firmware as of Nov 2019. However, the ambient temperature of the test environment was not given. It seems safe to assume that additional heat sinks and/or cooling fans were not used, as they are not shown in the graphics or discussed in the text.
The chart in the link in your question was taken from these tests. You can read some details on how they did their testing there. You can also read of similar testing done by an independent group (Tom's Hardware), and notice that the results are in reasonably close agreement. The Tom's Hardware test report indicates the ambient test environment was 24°C, and no mention was made of additional cooling.
You can also read the Power Supply Requirements published by "The Organization". Reading that, you will see that the "Recommended PSU Capacity" for the 4B is 3.0A - but also note that 1.2A are "set aside" for USB peripherals. This leaves 1.8A - approximately 9W for the RPi itself. That 9W figure is consistent with the measurements taken during testing, and it seems a prudent recommendation for most usage.
With respect to what "someone on the web" recommends, I can only offer this: With additional cooling (fans, heatsinks) and a lower ambient, it seems likely that the RPi could consume more power than in the tests referenced above - perhaps significantly more. Also, some will choose peripherals that may push the limit. But unless your usage case is extreme, the 3A recommendation seems sound - even conservative for many applications. If your usage case demands more power, I'll leave it to you to shop for a higher-powered adapter, but certainly they are available: for example Apple's Macbook USB-C charger is rated at 96W. And it's easy enough to assemble a 5V supply with very high amperage ratings.
In closing, if you're concerned about having enough power, consider purchasing a "USB Power meter". Plug one end of it into your RPi's USB-C input, and connect your USB cable to the other. This will tell you two things:
if your cable or your supply are sub-standard, the voltage may be less than the minimum (~4.8V)
you can monitor how much current your RPi and peripherals are drawing.