actually dont merge this yet cos working on another pr that might indirectly fix this :)

nevermind the other pr doesnt fix it so this can be merged when ready :)

Does this emit correct and easy to understand errors if the value it larger than 5 bits?
shamt may only be a 5 bit value from 0-31.
You are allowing constants up to 16 bits.

the errors will be the same as the imemdiate because (0..32).contains(n) is still checked

yeah they are :) same as immediates

a = 32
sll $0, $0, a
error: failed to compile `/dev/stdin`
 --> /dev/stdin:2:1
  |
2 | sll $0, $0, a
  | ^^^^^^^^^^^^^ instruction `sll` exists but was given incorrect arguments
tip: valid formats for `sll`:
  - sll $Rd, $Rt, shift
  - sll $Rd, shift
  - sll $Rd, i16, shift
  - sll $Rd, u16, shift
  - sll $Rd, i32, shift
  - sll $Rd, u32, shift

sidenote maybe it should be more clear to explicitly say something like ... shift ([0,32)) in the error because currently it feels a bit misleading. something as simple as this maybe?

-ArgumentType::Shamt => write!(f, "shift"),
+ArgumentType::Shamt => write!(f, "shift ([0,31])"),

edit: in #307 better errors are supported to be explicit about the 0..32 range instead of being a generic incorrect argument error

yeah k this entire pr is indirectly eclipsed by #307 . not sure if it should bother being merged?