Look at the microstrip lines of HDMI LVDS outputs. Two ways are possible for EMI to reach antenna - by air (outside of a PCB) or via power planes and power circuitry.
If some geeks measured emanating emissions, it seems more likely to be the former. In that case, it should be easier to fix it - just stick a paper with an aluminium foil on the top of these links. My favourite chocolate might help ;-).
Be aware that aluminium is conductive, don't destroy your Pi by shorting it. The best shielding effect is archieved if the edges are connected to earth shielding. On one side, it may be WiFi cage, on the other shields of HDMI connectors and on the third side perhaps the lid of a BCM chip itself. So a sandwich made of a thick paper (about 0.5mm) with an aluminium foil on top of it sticked to the PCB, touching on edges to RPI's grounding might help better than a software solution.
Software solution - downclock pixel rate a bit and shorten timing of blanking interval to keep the same framerate. For X server, it'd mean to create a new section, possibly dumped from a working configuration and slightly modded.
Or if you have a sticky metal tape (and I mean metal, not metal-like looking) - just stick it over HDMI differential pairs from the corner of a BCM chip to the HDMI ports.
With regard to CooliPi 4B heatsink - it may need an EMI gasket between the WiFi cage and a heatsink itself. Because it's so close to the Pi, it may already act as a shielding, albeit not with perfect attenuation. I've ordered some and will post results. Spectral analysers are available. Only to find some time :-(