Use scipy rotation matrices instead of by hand transformation

[StFroese]

I think this would be easier to read if we use something like
R.from_rotvec(vec).as_matrix() from scipy.

ctapipe/ctapipe/coordinates/ground_frames.py

Lines 112 to 124 in 5de39ce

	cos_z = sin(altitude)
	sin_z = cos(altitude)
	cos_az = cos(azimuth)
	sin_az = sin(azimuth)
	trans = np.array(
	[
	[cos_z * cos_az, -cos_z * sin_az, -sin_z],
	[sin_az, cos_az, np.zeros_like(sin_z)],
	[sin_z * cos_az, -sin_z * sin_az, cos_z],
	],
	dtype=np.float64,
	)

[maxnoe]

Does that support broadcasting many angles? That wa sthe problem the last time I tried to simplify that code.

[StFroese]

The documentation says: rotvec : array_like, shape (N, 3) or (3,)
matrix : ndarray, shape (3, 3) or (N, 3, 3)
Should work for many angles

[maxnoe]

Rotation.from_euler('zy', [az.to_value(u.rad), zen.to_value(u.rad)]).as_matrix()

[StFroese]

But has to be -zen because of Corsika coords