Kernel/aarch64: Don't set multiboot_modules to an empty array on-stack

[supercomputer7]

Since multiboot_modules_count is set to 0, we can safely set the multiboot_modules pointer to 0 (null pointer), as we don't use multiboot on aarch64 anyway.

[nico]

Could maybe say "fixed the aarch64 build after $commit_hash" in the commit description

[supercomputer7]

Could maybe say "fixed the aarch64 build after $commit_hash" in the commit description

I am not 100% sure which commit led to this. Is this my commit (897c4e5) in which I baked the prekernel and kernel together for x86?

[ADKaster]

Even if we go back before that PR, we were doing this awkward memcpy from nullptr of 0 * sizeof(mutliboot_module_t) bytes. I'm not sure I understand either why this is suddenly broke.

I just tested latest master with a GCC build and it seems to work fine without this patch? Who was actually broken by this? was it @BertalanD ?

[BertalanD]

Yes, it was me.

897c4e5 removed the multiboot_modules variable without also removing the assignment to it in the ARCH(AARCH64) code path.

https://discord.com/channels/830522505605283862/897113928168013855/1101503142375456799

[ADKaster]

Since we have since reverted 897c4e5 because of #18558, I think it's fine to merge this without pointing out that it fixes a secondary regression from it. 🤷