Better establish type conversion between numpy and C++ types

[vepadulano]

Explain what you would like to see improved and how.

There are a few places where we need to convert numpy dtypes to C++ types, for example when adopting buffers of numpy arrays into RVecs. There are at least the following two conversion dictionaries:

1. root/bindings/pyroot/pythonizations/python/ROOT/_pythonization/_rvec.py

   Line 64 in 65a0e12

   _array_interface_dtype_map = {

2. root/bindings/pyroot/pythonizations/python/ROOT/_pythonization/_rvec.py

   Line 78 in 65a0e12

   cpptypes = {"i2": "Short_t", "u2": "UShort_t", "i4": "int", "u4": "unsigned int", "i8": "Long64_t", "u8": "ULong64_t", "f4": "float", "f8": "double", "b1": "bool"}

Which do not contain currently the same set of semantic key-value pairs. Also, sometimes a builtin C++ type is used, sometimes a ROOT typedef is used, notably in the treatment of integer types. It would be beneficial to establish a set of all the types allowed for conversion between numpy and C++ types and resolve the inconsistencies. My preference would also be to avoid using ROOT typedefs and switch to C++ standard integer types. I expect this may lead to issues that will need to be investigated, see #18087

ROOT version

Any

Installation method

Any

Operating system

Any

Additional context

No response