is there any particular reason why those instructions use memory instead of register/imm?

load instruction can load 24 or 32 bits, + 16 of size. or 32/16 is used with 0x66.
why its not like ax hold size, and dx hold 16 bit address (or edx 32bit if l*dt used with 0x66)?

same s*dt, why do i have to create memory just to load this shit?
im for sure gona copy it to register in next instruction...

why intel made that in this way?

l*dt would lose capability of loading 24 bits, but cmon why bother with this. use 0x66 if u need even 32 bits.

It is too late to be asking about why Intel did certain things. It is what it is and is can't be changed now.