allow dynamic function creation for dtype copy

[JoranAngevaare]

What is the problem / what does the code in this PR do
Add a function to more easily numbafy the copy from dtype to dtype. Also see XENONnT/straxen#353

Can you briefly describe how it works?
Numba magic 🧙

Rather than writing each time a cumbersome function that has to get the dtype write, why not code this. The annoying thing about numba is that it does not allow you to use strings to loop over. This is a workaround.

Can you give a minimal working example (or illustrate with a figure)?

r = np.ones(1000, dtype=strax.raw_record_dtype()
def test_raw_to_records(r):
    buffer = np.zeros(len(r), r.dtype)
    strax.copy_to_buffer(r, buffer, "_test_r_to_buffer")
    if len(r):
        assert np.all(buffer == r)

Special thanks to
Daniel for helping writing this up nicely and Jim Pivarski for showing such functionality in numba.

[WenzDaniel]

Thanks a lot Joran for this awesome idea. It makes copying data much less tedious.