CodeGen: Optimize arithmetics for basic identities

[zeux]

This change folds:

a * 1 => a
a / 1 => a
a * -1 => -a
a / -1 => -a
a * 2 => a + a
a / 2^k => a * 2^-k
a - 0 => a
a + (-0) => a

Note that the following folds are all invalid:

a + 0 => a (breaks for negative zero)
a - (-0) => a (breaks for negative zero)
a - a => 0 (breaks for Inf/NaN)
0 - a => -a (breaks for negative zero)

Various cases of UNM_NUM could be optimized (eg (-a) * (-b) = a * b),
but that doesn't happen in benchmarks.

While it would be possible to also fold inverse multiplications (k * v),
these do not happen in benchmarks and rarely happen in bytecode due
to type based optimizations. Maybe this can be improved with some sort of
IR canonicalization in the future if necessary.

I've considered moving some of these, like division strength reduction,
to IR translation (as this is where POW is lowered presently) but it didn't
seem better one way or the other.

This change improves performance on some benchmarks, e.g. trig and voxelgen,
and should be a strict uplift as it never generates more instructions or longer
latency chains. On Apple M2, without division->multiplication optimization, both
benchmarks see 0.1-0.2% uplift. Division optimization makes trig 3% faster; I expect
the gains on X64 will be more muted, but on Apple this seems to allow loop iterations
to overlap better by removing the division bottleneck.

[zeux]

After testing this on math_map equivalent, I've went ahead and added a-0/a+(-0) folding after all. With it, this code:

local function math_map(x: number, inmin: number, inmax: number, outmin: number, outmax: number): number
	return outmin + (x - inmin) * (outmax - outmin) / (inmax - inmin)
end

local function lerpopt(a, b, t)
	return math_map(t, 0, 1, a, b)
end

gets compiled into this:

 ldr         w17,[x25,#44]
 cmp         w17,#3
 b.ne        .L48
 ldr         d31,[x25,#32]
 str         d31,[x25,#96]
 movz        w17,#3
 str         w17,[x25,#108]
.L49:
 ldr         w17,[x25,#28]
 cmp         w17,#3
 b.ne        .L50
 ldr         w17,[x25,#12]
 cmp         w17,#3
 b.ne        .L50
 ldr         d30,[x25,#16]
 ldr         d29,[x25]
 fsub        d30,d30,d29
 fmul        d31,d31,d30
 fadd        d29,d29,d31
 str         d29,[x25,#48]
 movz        w17,#3
 str         w17,[x25,#60]
 ldr         x0,[x21,#3296]
 cbnz        x0,.L51
.L52:
 ldr         q0,[x25,#48]
 str         q0,[x25,#-16]
 mov         x1,x25
 movz        w2,#1
 b           .L7

... which is almost good, short of an extra dead store in the first basic block that should have been elided; explicit type annotations on lerpopt help fix that though.

[vegorov-rbx]

Thank you.