...assuming the GPU has the horsepower to drive it at > HDMI 2.0 (or > DP 1.2) rates. Unless Apple have artificially knobbled the M1 Pro MBPs and/or (OK, that wouldn't be the shock of the century) to only support 2 displays and/or not support HDMI 2.1, they don't have DisplayPort bandwidth to spare.
The M1 chips only support DisplayPort natively. All of the PHYs are DisplayPort 1.4a (or eDP 1.4b) and capable of HBR3 signaling (8.1 Gbit/s per lane), Variable Refresh Rate (VRR), Forward Error Correction (FEC), and Display Stream Compression (DSC). VRR is only available when running macOS 12 Monterey or later. HDMI 2.0 TMDS links are enabled by the same protocol converter chip that Apple uses on at least a dozen other products they currently sell. There is currently no macOS support whatsoever for HDMI 2.1 FRL, including protocol converters. Although macOS is aware of DisplayPort MST, it is only supported for mirroring or on specific displays that use MST to drive tiled panels. There is no shortage of DisplayPort bandwidth, unless you consider the lack of DisplayPort 2.0 UHBR signaling rates a problem.
The limitations on rendering pipelines and number of display output streams are pretty obviously tied to the size / capabilities of the GPUs. The M1 can only support two display output streams, one at up to 5120 x 4096 (2560 x 2048 HiDPI) and the other at up to 6144 x 4096 (3072 x 2048 HiDPI), and up to 10 bits per component (32 bits per pixel, ARGB2101010). The M1 Max can handle 5 streams at up to 7680 x 4096 (3840 x 2048 HiDPI) resolution. The M1 Pro appears to be limited to 3 display output streams, but I don't have the ioreg output for one of those handy at the moment. You do realize that scaling from 7-8 GPU cores + 68.3 GB/s memory bandwidth, to 14-16 GPU cores + 204.8 GB/s memory bandwidth, to 24-32 GPU cores + 409.6 GB/s memory bandwidth has performance implications?
Interesting. I was pretty sure that Apple specifically said that the M1 pro had an extra TB controller over the M1 - but I see the printed press release just vaguely says "extra TB controllers".
I won't argue with your IC component identification skills - it certainly looks plausible - although I don't get why the "oddball in the middle" would have two DisplayPort sections...
Also, it's impossible to know whether all of those supposed TB and PCIe controllers can be enabled simultaneously and/or if they are all equal in terms of connectivity to the GPU and CPU. E.g. one of them could be a "runt" just for supplying USB 3.1g1 and HDMI. I'm also thinking of the Intel chipsets which had a bunch of "universal I/O" lanes that could be configured as either USB, general PCIe, SSD-optimised PCIe etc. but only in certain fixed permutations.
...so, maybe the mains-powered iMac will have 4xTB with an extra TB controller enabled at the expense of slightly higher power consumption... or the Mac Mini Pro will have 5 TB ports because it doesn't need to support an internal display... or the iMac Pro XDR will use that "oddball" controller for an 8k display.... I.e. the M1 Pro/Max chips in MacBook Pro, iMac Pro, Mac Mini Pro (...and future Mac Pro?) might be the same physical chips but with different permutations of modules enabled at build time.
Of course, it would be nice if Apple would share this info so we didn't have to speculate, but they do like their secrets.
Many of the Intel discrete Thunderbolt controllers also included a "display side port". It allows you to have a dedicated display output port (DP, HDMI, DVI, whatever) as well as a full-featured Thunderbolt port
with the display signal muxed to both so you can just route to whichever one is in use.
edit: Bad explanation. The Thunderbolt controller has two DisplayPort inputs that can be routed to any of four downstream destinations: either of the two DP-in protocol adapters used for DisplayPort protocol tunneling, the native DP PHY for USB Type-C DisplayPort Alternate Mode on the Thunderbolt / USB4 port, or a native DP PHY for a dedicated display output port.
It looks like Apple is using Synopsys USB4 IP all around for the USB xHCIs. There doesn't appear to be any standalone USB 3.2 Gen1x1/Gen2x1 stuff anywhere, hence the use of discrete controllers from Fresco Logic on the Mac mini and ASMedia on the 24-inch iMac so they didn't have to burn one of the two available Thunderbolt / USB4 ports.
