fix: replace np.asarray by np.array in dataclass constructors

[ktro2828]

What

This PR replaces np.asarray in converters in each dataclass into np.array.
Because np.asarray doesn't create a copy of the input if it is already ndarray, the original values are changed unexpectedly if a destructive method is called, such as Box3D.translate(...).

Experiment

Test code is as follows:

from t4_devkit import Tier4

t4 = Tier4("annotation", <DATA_ROOT>, verbose=False)

print(t4.sample_annotation[0].translation)

# Destructive methods, which are Box3D.translate(...) and Box3D.rotate(...), are invoked under the hood.
t4.get_sample_data(t4.sample_data[0].token)  

print(t4.sample_annotation[0].translation)

Before

Then the output shows the original has been changed.

[8.9548508e+04 4.2286165e+04 7.9690000e+00]
[-6.61751375 -4.48131001  1.01779332]

After

As of this PR, there is no change between before and after calling destructive methods.

[8.9548508e+04 4.2286165e+04 7.9690000e+00]
[8.9548508e+04 4.2286165e+04 7.9690000e+00]

[github-actions]

☂️ Python Coverage

current status: ❌

Overall Coverage

Lines	Covered	Coverage	Threshold	Status
2106	1428	68%	50%	🟢

New Files

No new covered files...

Modified Files

File	Coverage	Status
t4_devkit/common/converter.py	89%	🟢
t4_devkit/dataclass/box.py	76%	🔴
t4_devkit/dataclass/pointcloud.py	40%	🔴
t4_devkit/dataclass/roi.py	94%	🟢
t4_devkit/dataclass/shape.py	94%	🟢
t4_devkit/dataclass/trajectory.py	93%	🟢
t4_devkit/dataclass/transform.py	76%	🔴
t4_devkit/schema/tables/calibrated_sensor.py	95%	🟢
t4_devkit/schema/tables/ego_pose.py	95%	🟢
t4_devkit/schema/tables/keypoint.py	94%	🟢
t4_devkit/schema/tables/sample_annotation.py	97%	🟢
t4_devkit/schema/tables/vehicle_state.py	100%	🟢
TOTAL	87%	🔴

updated for commit: 251c10d by action🐍