ASM Optimization for "null operations"

[xunilrj]

If we compile the code below with forc build --path {root} --release --ir final.

fn small_array() -> [bool; 5] {
    [true; 5]
}

We see the following ASM:

move $$locbase $sp            ; save locals base register for function small_array_repeat_11
move $r0 $$arg0               ; save argument 0 (__ret_value)
move $r1 $$reta               ; save return address
movi $r2 i1                   ; get array element size
mul  $r2 $zero $r2            ; get offset to array element
add  $r2 $r0 $r2              ; add array element offset to array base
sb   $r2 $one i0              ; store byte
movi $r2 i1                   ; get array element size
mul  $r2 $one $r2             ; get offset to array element
add  $r2 $r0 $r2              ; add array element offset to array base
sb   $r2 $one i0              ; store byte
movi $r2 i1                   ; get array element size
movi $r3 i2                   ; initialize constant into register
mul  $r2 $r3 $r2              ; get offset to array element
add  $r2 $r0 $r2              ; add array element offset to array base
sb   $r2 $one i0              ; store byte
movi $r2 i1                   ; get array element size
movi $r3 i3                   ; initialize constant into register
mul  $r2 $r3 $r2              ; get offset to array element
add  $r2 $r0 $r2              ; add array element offset to array base
sb   $r2 $one i0              ; store byte
movi $r2 i1                   ; get array element size
movi $r3 i4                   ; initialize constant into register
mul  $r2 $r3 $r2              ; get offset to array element
add  $r2 $r0 $r2              ; add array element offset to array base
sb   $r2 $one i0              ; store byte
move $$retv $r0               ; set return value
move $$reta $r1               ; restore return address
poph i524288                  ; restore registers 40..64
popl i15                      ; restore registers 16..40
jmp $$reta                    ; return from call

We can see a lot of "null operations" such as mul by zero and others.
For example:

movi $r2 i1                   ; get array element size
mul  $r2 $zero $r2            ; get offset to array element
    ; $r2 is guaranteed to be one. We could use
    ; $one and kill the movi above
    ;
    ; but we are multiplying by zero
    ; so we can remove this mul and the movi
    ; and just use $zero
add  $r2 $r0 $r2              ; add array element offset to array base
    ; we would replace $r2 with $zero here
    ; but sum by zero is useless
    ; we know that $r2 = $r0
    ; remove this add
sb   $r2 $one i0              ; store byte
    ; replace $r2 by $r0 here

We already have these kind optimizations, but it seems there is still space to improve them.