models: Explicitly set Z&P 2008 data file encoding

Fixes some corner cases where the utf-8 format of the Zhang and Paxton
2008 coefficient data file is not correctly detected and
`numpy.genfromtxt()` fails on the unicode characters.

Increases the `numpy` minimum version since the `encoding` keyword was
introduced in version 1.14.0.

setup.py

@@ -78,7 +78,7 @@ def find_meta(meta, *path):
 		},
 		include_package_data=True,
 		install_requires=[
-			"numpy>=1.13.0",
+			"numpy>=1.14.0",
 			"scipy>=0.9",  # LinearNDInterpolator
 		],
 		extras_require=extras_require,

src/eppaurora/models/zhangpaxton2008.py

@@ -65,6 +65,7 @@ def read_zp2008_coeffs(file=None, nf=6, nKp=len(KP_BINC)):
 		file or COEFF_PATH,
 		delimiter=" ",
 		dtype=None,
+		encoding="utf-8",
 		names=["name"] + COEFF_NAMES,
 	)
 	# number of coefficients per Kp bin