tpm2: Filter bad input values to avoid underflow in FindNthSetBit (Coverity)

Address the following Coverity complaint (1550494) by filtering out bad
input values:

  "Expression i--, which is equal to 65535, where i is known to be equal
   to 0, underflows the type that receives it, an unsigned integer 16 bits
   wide."

aSize is typcially 2048 and n is always >= 1 per the input parameter.
Therefore no side-effects are expected from this filter.

Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>

src/tpm2/crypto/openssl/CryptPrimeSieve.c

@@ -215,6 +215,10 @@ FindNthSetBit(
     int          retValue;
     UINT32       sum = 0;
     BYTE         sel;
+
+    if (n < 1 || aSize < 1)			// libtpms added begin: Coverity 1550494
+        return -1;				// libtpms end
+
     //find the bit
     for(i = 0; (i < (int)aSize) && (sum < n); i++)
 	sum += BitsInByte(a[i]);