How data processes in microprocesser ??

Which would not relieve a conforming C compiler from jumping through whatever hoops needed to perform an actual shift by 10.

I agree with Simon, most likely this was intended to provide some sort of clarity/documentation.

A second, and lesser, possibility is a very poor compiler that optimized the double shift form better, perhaps by generating a 2 bit shift and then a byte move, as opposed to a (perhaps very) slow 10 bit shift.

Now, it just gets in the way of readability, and sometimes messes up the optimizer (when it's a more complicated "clue" than that one).

But it's a compile time known constant, so surely the compiler would just evaluate the expression at compile time and just place the final result in the generated code.

Back in the 1990s (your twenty years ago) I was writing LL(1) recursive descent based compilers and that was normal behaviour even in the simple compilers I was writing then.

Perhaps the original author wanted to make clear what bit pattern lives in the top 8 bits of the opcode.

Simon.

I was using some pretty simple-minded compilers. And some good ones. I recall spending a lot more time dodging compiler oddities than I do now.

That could be, too. Certainly with a modern compiler that's the only excuse.