Fix memory leak when creating MPIs for ECGroup

[arckoor]

In #5166 (comment) copilot called out that what I was doing in the python code would leak the MPI initialized by __init__. This is the same thing ECGroup is already doing, so fix it.
To be honest, I don't love this fix, I've omitted the | c_void_p from the MPILike type hint, because it's used in other places and I don't think encouraging people to pass a magic pointer that hopefully is an MPI to something like PrivateKey.load_ecdsa(curve: str, x: MPILike) is a good idea.
The MPI class is very unfortunate, all the other classes basically just assign a pointer to self.__obj and let whatever else do the proper initialization, MPI is seemingly the only one that does extra stuff in its constructor.
I considered just skipping __init__ and implementing / calling __new__ directly, but it appears that is frowned upon in the python community (which is probably a good thing).

[Copilot]

Pull request overview

This pull request fixes memory leaks in the ECGroup class methods that retrieve MPI (multi-precision integer) values. The issue was that calling MPI() initializes an MPI object by allocating memory via botan_mp_init, but then the FFI functions like botan_ec_group_get_p would overwrite the pointer with a newly allocated MPI, causing the first allocation to leak.

Changes:

• Modified MPI.init to accept c_void_p and skip initialization when a raw pointer is provided
• Updated all six ECGroup getter methods (get_p, get_a, get_b, get_g_x, get_g_y, get_order) to use the new pattern
• Updated version comment from 3.10.0 to 3.11.0

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[coveralls]

coverage: 90.332% (-0.003%) from 90.335%
when pulling a92e237 on arckoor:python-ec-group-memleak
into ef0dcf4 on randombit:master.