json: improve parsing and serializing of integers

Add support for parsing and serializing of following integer types:
'int8_t', 'uint8_t', 'int16_t', 'uint16_t' and 'uint32_t'.

The generic integer token JSON_TOK_INT and JSON_TOK_UINT, in combination
with the field size (set by JSON_OBJ_DESCR_PRIM) allows to parse different
integer types, for example:

struct foo {
  int64_t i64;
  uint32_t u32;
  int16_t i16;
  uint8_t u8;
};

struct json_obj_descr foo_descr[] = {
  JSON_OBJ_DESCR_PRIM(struct foo, i64, JSON_TOK_INT),
  JSON_OBJ_DESCR_PRIM(struct foo, u32, JSON_TOK_UINT),
  JSON_OBJ_DESCR_PRIM(struct foo, i16, JSON_TOK_INT),
  JSON_OBJ_DESCR_PRIM(struct foo, u8, JSON_TOK_UINT),
};

These tokens also support parsing and serializing enums:

enum unsigned_enum { UA=0, UB=1, UC=2 };
enum signed_enum { SA=-1, SB=0, SC=1 };

struct foo {
  enum unsigned_enum u;
  enum signed_enum s;
};

struct json_obj_descr foo_descr[] = {
  JSON_OBJ_DESCR_PRIM(struct foo, u, JSON_TOK_UINT),
  JSON_OBJ_DESCR_PRIM(struct foo, s, JSON_TOK_INT),
};

Signed-off-by: Christoph Winklhofer <cj.winklhofer@gmail.com>