Fix a few issues with the C generator (part 7)

[eafer]

Another new warning I've noticed when building our api with the latest version:

/root/avh-api/c/api/api/ArmAPI.c: In function ‘ArmAPI_v1CreateImage’:
/root/avh-api/c/api/api/ArmAPI.c:1346:32: error: assignment to ‘int’ from ‘void *’ makes integer from pointer without a cast [-Werror=int-conversion]
 1346 |         valueForm_encapsulated = calloc(1,MAX_NUMBER_LENGTH);
      |                                ^
/root/avh-api/c/api/api/ArmAPI.c:1347:18: error: passing argument 1 of ‘snprintf’ makes pointer from integer without a cast [-Werror=int-conversion]
 1347 |         snprintf(valueForm_encapsulated, MAX_NUMBER_LENGTH, "%d", *encapsulated);
      |                  ^~~~~~~~~~~~~~~~~~~~~~
      |                  |
      |                  int
In file included from /root/avh-api/c/api/api/ArmAPI.c:2:
/usr/include/stdio.h:378:39: note: expected ‘char * restrict’ but argument is of type ‘int’
  378 | extern int snprintf (char *__restrict __s, size_t __maxlen,
      |                      ~~~~~~~~~~~~~~~~~^~~

This is not actually a regression, the code was always broken but recent changes made the warning show up. This pull request does the following:

1. Update the test schema to cover this broken case.
2. Change the compiler flags to ban implicit int to pointer conversions, so that automated testing will not miss these warnings.
3. Fix the actual bug, changing the type of valueForm to char *.

There are other issues with formData input. I think valueForm should become char * for all types, not just bool. But this involves bigger changes, so I might look into it at some later time.

[eafer]

@wing328 @ityuhui @zhemant @michelealbano

[eafer]

Great! Final pull request here: #20378.