arm64: dts: apple: Add PCIe nodes for t6031

[yuyuyureka]

Adds the pcie0 node on die0 of all M3 Max / Ultra chips (t6031/t6032/t6034), as
well as pwren-gpios for WiFi and Bluetooth of the MacBook Pros using these
chips.

The M3 Ultra t6032, which is part of this family of devices, theoretically
has a pcie1 node, but it's not directly used for anything except nvme, which
is already present in the device tree.

The pwren- and reset-gpios have been sourced from Apple's device trees, and are
confirmed on the MacBook Pro 14", 14-cores, 2023 (j514m).

Once the t6030 pcie0 node is added, we can move the WiFi / Bluetooth parts into
t603x-j514-j516.dtsi, as the pwren-gpios are the same across all
M3 / M3 Pro / M3 Max MacBook Pros.

[WhatAmISupposedToPutHere]

should have 4 pm domains, right?

[yuyuyureka]

I'm not sure? ps_apcie_sys_gp depends on ps_apcie_gp, ps_afnc3_lw0, ps_afnc3_ls, ps_afnc3_ioa (transitively).

[WhatAmISupposedToPutHere]

if i am reading the adt right, it also wants ps_apcie_sys_ge, ps_apcie_sys_st, and ps_apcie_sys_st1

[jannau]

I'm not convinced that ps_apcie_sys_st and ps_apcie_sys_st1 are needed. nvme power-states also has dependencies on apcie_*{ge,gp} but worked without them. ps_apcie_sys_ge is missing

[jannau]

pwren-gpios need top be in a separate commit as that's not upstreamable. Adding the pcie ports to the device .dts is also easier to manage as separate commit.

[jannau]

partially merged to asahi-wip. I added ps_apcie_sys_ge to the power-domains but skipped all device changes. These can now be done in t603x-j514-j516.dtsi as t6030.dtsi has PCIe nodes as well.