Add support for arrays in CoreRangeSet (#14967)

### Ticket
- #14962

### Problem description
For the upcoming matmul1d op that supports inputs sharded on arbitrary
cores, the ordering of the cores must be accessible in the program
factory. This is so that we can pick a ring such that the NOC directions
are optimized according to the following:
<img width="1016" alt="image"
src="https://github.com/user-attachments/assets/730c72e3-f79c-4b3d-a281-50c19db252ba">


### What's changed
- Added support to pass in arrays to the CoreRangeSet object.

ttnn/cpp/pybind11/tensor.cpp

@@ -256,6 +256,7 @@ void tensor_mem_config_module(py::module& m_tensor) {
 
     auto pyCoreRangeSet = static_cast<py::class_<CoreRangeSet>>(m_tensor.attr("CoreRangeSet"));
     pyCoreRangeSet.def(py::init<>([](const std::set<CoreRange>& core_ranges) { return CoreRangeSet(core_ranges); }))
+        .def(py::init<>([](const std::vector<CoreRange>& core_ranges) { return CoreRangeSet(tt::stl::Span<const CoreRange>(core_ranges)); }))
         .def(
             "bounding_box",
             &CoreRangeSet::bounding_box,