[Coverity] Medium issues with GPU plugin (#27686)

### Details:
- Ignoring the following CIDs since the status is already fixed:
1548290, 1546891, 1546890, 1546888, 1546867, 1546862, 1546023, 1546007,
- Ignoring the following CIDs since it is third party code: 1548253,
1545465
- Ignoring 1548233, believe it's false positive.
- Ignoring 1549003, per reviewers feedback it is false positive. Updated
Coverity status.
- This PR, fixes: 1548264

### Tickets:
 - [CVS-145083](https://jira.devtools.intel.com/browse/CVS-145083)

---------

Co-authored-by: Pavel Durandin <[email protected]>

src/plugins/intel_gpu/src/graph/primitive_inst.cpp

@@ -2028,7 +2028,7 @@ primitive_inst::primitive_inst(network & network, program_node const& node, bool
     , _inputs_memory_count(node.get_inputs_count())
     , _outputs_memory_count(node.get_outputs_count())
     , _fused_mem_count(node.get_fused_inputs_count())
-    , _fused_mem_offset((_fused_mem_count > 0 && node.get_first_fused_dep_idx() > 0) ? node.get_first_fused_dep_idx() : 0)
+    , _fused_mem_offset((_fused_mem_count > 0 && node.get_first_fused_dep_idx() > 0) ? static_cast<uint64_t>(node.get_first_fused_dep_idx()) : 0)
     , _can_be_optimized(node.can_be_optimized())
     , _can_share_buffer(node.can_share_buffer())
     , _is_constant(node.is_constant())

src/plugins/intel_gpu/src/graph/program_node.cpp

@@ -1505,7 +1505,7 @@ void program_node::create_onednn_primitive_attributes(
                     oc_dim = static_cast<int>(desc.output_layout.get_partial_shape()[1].get_max_length());
                 else
                     oc_dim = static_cast<int>(desc.output_layout.get_tensor().feature.size());
-                post_ops.append_prelu(1 << std::max(0, oc_dim));
+                post_ops.append_prelu(1 << static_cast<unsigned>(std::max(0, oc_dim)));
                 update_onednn_post_op_list(onednn_post_op_type::binary_relu, dep_idx);
             } else if (fused_desc->activation_function == cldnn::activation_func::hard_sigmoid) {
                 post_ops.append_eltwise(dnnl::algorithm::eltwise_hardsigmoid, fused_desc->additional_params.a, fused_desc->additional_params.b);

src/plugins/intel_gpu/src/plugin/transformations/dynamic_quantize_fully_connected.cpp

@@ -18,7 +18,7 @@ namespace ov {
 namespace intel_gpu {
 
 DynamicQuantizeFullyConnected::DynamicQuantizeFullyConnected(uint64_t group_size)
-    : ov::pass::MatcherPass() {  // Explicitly call the base class constructor
+    : ov::pass::MatcherPass() {
     GPU_DEBUG_GET_INSTANCE(debug_config);
     using namespace ov::pass::pattern;