Fixes UTF8 output length by accounting for variable header size

[alex-miller-0]

cb0r was failing to decode data whose key was >23 bytes. This is because 23 is the largest value which does not include a length prefix, per the cbor spec.

Here is an example of the issue:

#include "cb0r.h"
#include "string.h"
#include "stdio.h"
#define NOT_TOO_LONG_KEY "HelloIHaveAVeryLongName"
#define TOO_LONG_KEY "HelloIHaveAVeryLongName!"
int main() {
  // { "HelloIHaveAVeryLongName": 0 }
  uint8_t notTooLong[26] = {
    0xa1, 0x77, 0x48, 0x65, 0x6c, 0x6c, 0x6f, 0x49, 
    0x48, 0x61, 0x76, 0x65, 0x41, 0x56, 0x65, 0x72, 
    0x79, 0x4c, 0x6f, 0x6e, 0x67, 0x4e, 0x61, 0x6d,
    0x65, 0x00, 
  };
  // { "HelloIHaveAVeryLongName!": 0 }
  uint8_t tooLong[28] = {
    0xa1, 0x78, 0x18, 0x48, 0x65, 0x6c, 0x6c, 0x6f, 
    0x49, 0x48, 0x61, 0x76, 0x65, 0x41, 0x56, 0x65, 
    0x72, 0x79, 0x4c, 0x6f, 0x6e, 0x67, 0x4e, 0x61, 
    0x6d, 0x65, 0x21, 0x00
  };
  cb0r_s result1 = {0};
  cb0r_s data1 = {0};
  if (!cb0r_read(notTooLong, sizeof(notTooLong), &data1))
    return -1;
  bool found = cb0r_find(&data1, CB0R_UTF8, 23, (uint8_t*)NOT_TOO_LONG_KEY, &result1);
  printf("Found not too long key: %d\n\r", found);
  cb0r_s result2 = {0};
  cb0r_s data2 = {0};
  if (!cb0r_read(tooLong, sizeof(tooLong), &data2))
    return -1;
  // FUNNY ENOUGH THIS PASSES IF I USE 25 INSTEAD OF 24
  bool found2 = cb0r_find(&data2, CB0R_UTF8, 24, (uint8_t*)TOO_LONG_KEY, &result2);
  printf("Found too long key: %d", found2);
  return 0;
}

Note the prefix of the 23-character key is 0x77, while that of the 24-character key is 0x78, 0x18, where 0x18 indicates the length of 24.

Failing to account for this extra byte led to result->length being invalid, which led to a decoding failure. Simply capturing this header size results in proper decoding.

Note that we only needed to fix this for UTF8 types and while we have smoke-tested a large range of types and sizes, there are some types we do not use so similar issues may be present in other areas of the code.

[letmaik]

Confirmed that this also fixes length of byte arrays, unsurpringly.