Fixed array struct types generation

[pixsperdavid]

What is the rationale behind generating a separate struct type wrapping each basic array in the C header? E.g. for a struct:

#[derive_ReprC)]
#[repr(C)]
pub struct Foo {
    array_a: [u8; 8],
    array_b: [u8; 32],
    array_c: [u16; 8]
}

I end up with:

typedef struct {
    uint8_t idx[8];
} uint8_8_array_t;

typedef struct {
    uint8_t idx[32];
} uint8_32_array_t;

typedef struct {
    uint16_t idx[8];
} uint16_8_array_t;

/** <No documentation available> */
typedef struct Foo {
    /** <No documentation available> */
    uint8_8_array_t array_a;

    /** <No documentation available> */
    uint8_32_array_t array_b;

    /** <No documentation available> */
    uint16_8_array_t array_c;
} Foo_t;

Which feels like a lot of extra noise to add the header, and additionally requires users to access data via the idx field which is not particularly ergonomic. Is there a good reason for doing this? I'm generating a header for public consumption (in an SDK) and I'm generally concerned about keeping things simple and readable. I would prefer to just use the arrays directly, eg:

/** <No documentation available> */
typedef struct Foo {
    /** <No documentation available> */
    uint8_t array_a[8];

    /** <No documentation available> */
    uint8_t array_b[32];

    /** <No documentation available> */
    uint16_t array_c[8];
} Foo_t;

Happy to draft a PR to enable the above generation behaviour as an option.