Address Numpy deprecation warning for out-of-bound integers in xli_unpack

[mrjonesolmedo]

What

Explicitly casts the result of xli_unpack to a np.int16 to avoid a deprecation warning. Follows the suggestion in the warning to maintain the old behavior, which is to follow C integer overflow logic.

Why

I started to see a Numpy deprecation warning:

./sierra-ecg-tools/pysierraecg/venv/lib/python3.8/site-packages/sierraecg/xli.py:55: DeprecationWarning: NumPy will stop allowing conversion of out-of-bound Python integers to integer arrays.  The conversion of 65400 to int16 will fail in the future.
For the old behavior, usually:
    np.array(value).astype(dtype)`
will give the desired result (the cast overflows).
  unpacked[i] = (((buffer[i] << 8) | buffer[len(unpacked) + i]) << 16) >> 16

which is explained in more detail in the Numpy 1.24.0 release notes.

To replicate:

import warnings
from src.sierraecg import read_file

filepath = "../examples/xml/1_04_01/2020-5-18_15-48-11.xml"

with warnings.catch_warnings():
    warnings.simplefilter("always")
    ecg = read_file(filepath)