[catalogue] Uses getBytes() instead of byteAt() to get blob from OCCI

Should be more efficient as it's a single copy operation instead of one
character at-a-time, but still requires 2 copies of the blob to convert
it from occi::Bytes to std::string (as it needs an intermediate char*
array).

rdbms/wrapper/OcciRset.cpp

@@ -130,9 +130,9 @@ std::string OcciRset::columnBlob(const std::string &colName) const {
   try {
     const int colIdx = m_colNameToIdx.getIdx(colName);
     auto raw = m_rset->getBytes(colIdx);
-    std::string ret;
-    for(unsigned i = 0; i < raw.length(); ++i) ret.push_back(raw.byteAt(i));
-    return ret;
+    std::unique_ptr<unsigned char[]> bytearray(new unsigned char[raw.length()]());
+    raw.getBytes(bytearray.get(), raw.length());
+    return std::string(reinterpret_cast<char*>(bytearray.get()), raw.length());
   } catch(exception::Exception &ne) {
     throw exception::Exception(std::string(__FUNCTION__) + " failed for SQL statement " + m_stmt.getSql() + ": " +
       ne.getMessage().str());