Possible big endian breakage in 2.4.1

[ondrejholy]

Describe the bug
On the stable-2.0 branch I see the following build warning:

/FreeRDP/libfreerdp/core/gateway/rdg.c: In function ‘rdg_handle_ntlm_challenge’:
/FreeRDP/libfreerdp/core/gateway/rdg.c:1218:65: warning: passing argument 4 of ‘crypto_base64_decode’ from incompatible pointer type [-Wincompatible-pointer-types]
 1218 |         crypto_base64_decode(token64, (int)len, &ntlmTokenData, &ntlmTokenLength);
      |                                                                 ^~~~~~~~~~~~~~~~
      |                                                                 |
      |                                                                 size_t * {aka long unsigned int *}
In file included from /FreeRDP/include/freerdp/crypto/tls.h:23,
                 from /FreeRDP/libfreerdp/core/gateway/rdg.h:33,
                 from /FreeRDP/libfreerdp/core/gateway/rdg.c:36:
/FreeRDP/include/freerdp/crypto/crypto.h:111:52: note: expected ‘int *’ but argument is of type ‘size_t *’ {aka ‘long unsigned int *’}
  111 |                                               int* res_length);
      |                                               ~~~~~^~~~~~~~~~

and the same for rpc_ncacn_http_recv_in_channel_response and rpc_ncacn_http_recv_out_channel_response. Although, it seems to be safe for little endian, I am pretty sure this might be a problem on big endian...

Environment:

• OS: Linux
• Version/Distribution: Fedora 35
• Architecture: amd64

Additional context
This was introduced by #7366 resp. f0a0683.

This is not a problem on master as crypto_base64_decode really uses size_t * there.

[akallabeth]

@ondrejholy all occurrences should be fixed now, closing.

[ondrejholy]

I've just verified that, thanks.