madsExperimentalBranch

This PR applies the following changes:

• relocate the four compression schemes (no compression, lz4, bitshuffle, zlib) into respective header (.hpp) and source (.cpp) files -- in order to do this, enums and structs that are used throughout the seedee project have been moved from seedee.cpp/h into a separate header file (seedeeStructs.h)
• the 'utility' functions (e.g., array_to_json, n_elements, get_shape, check_protocol) have been moved from the main seedee.cpp file into a separate utilities/utils.cpp file with respective header file
• error handling throughout the seedee library is done with a c++11 compliant header only implementation of std::expected (which is c++23) in utilities/expected.hpp via tl::expected. Whenever an error is detected (e.g. faulty json input or similar), then an seedee_Error enum rather than the return value (typically an int) is returned and propagates up through the function calls until it is handled in the get_buffer_size/serialize_ndarray/get_data_size/deserialize_ndarray functions through a return as int through its C interface and currently also as an console output.
• access to json fields is done through a check .contains("") follows by .at(""). If desired json field is non-existent, then an seedee_Error::INVALID_JSON is reported.
• const keywords are introduced whenever possible (possible some missed opportunities)
• the data type and byte order is now handled through enums rather than chars for improved clarity (enum values defined to be corresponding char values for easy conversion to python)
• rather than having a inheritance of the serializers, there now is a struct Serialzer in serializer.cpp/hpp that contains arrays of function pointers to each of the implementations of the four compressions schemes functions. These pointers are then called in the impl functions in seedee.cpp.