Incorporate type annotations including the array shape in nanoDFT IterationState class

[hatemhelal]

In PR #71 we added additional type annotations to the IterationState class. Most of the fields of this data class are now annotated with a combined data type and tensor shape annotation. For example:

FloatNxN = Float[Array, "N N"]

This issue is to rationalise the remaining fields that are currently without a documented shape:

grid_weights
input_floats
input_ints
diis_history