[BUG] `SIMD.store/load` for `DType.bool`

[rd4com]

Bug description

Hello, after implementing List.__contains__ for Scalar[DT], tests failed with DType.bool !

Minimal repro:

alias width = simdwidthof[DType.bool]()
var x = DTypePointer[DType.bool].alloc(width)
alias T = SIMD[DType.bool, width]
T.store(x, T(True))
print(T.load(x)) #32 elements True
x[0] = False
print(T.load(x)) #8 firsts elements False
x.free()

Note: it seem to work correctly with others DType

Important: if we initialized with T.store(x, T(False)) instead, and do x[0] = True, it seem to work correctly, but it's not: assigning to offset 1 results in item 8 being changed

This test shows different results:

for i in range(width): 
    print(x[i])`        # see first four bytes

Possible fix: It seem to think that the size of the type is 8 bytes (debugging with DTypePointer.__setitem__ and __getitem__)

(ping to @laszlokindrat)

Steps to reproduce

Nothing special, the steps are in the description :+1:

System information

- Ubuntu 23.10
- mojo 2024.7.1105 (211527ca)
- modular 0.7.2 (d0adc668)