Techdebt refatorar modulo de adminclient do codex neodv 2150

[henrique-lh]

Description

With this pr:

• Add new admin module
• Improve code style
• Refactor python client to be the same entry point for all operations

Related issues

• Closes: https://linear.app/datarisk/issue/NEODV-2150/techdebt-refatorar-modulo-de-adminclient-do-codex
• https://linear.app/datarisk/issue/NEODV-2149/techdebt-refatorar-modulo-de-treino-do-codex

How to test it

• Set the API locally and change the BASE_URL in src/mlops_codex/constants.py
• Setup the codex locally

uv build
uv pip install -e .

Run the following script:

from mlops_codex.administrator import DatariskMLops
from mlops_codex.train import CustomTrain

client = DatariskMLops(
    email="<email>",
    password="<password>",
    tenant="<tenant>"
)

client.list_groups()

group_token = client.create_group(name="codexGroup", description="Codex Group")

print("refresh group token")

new_group_token = client.refresh_group_token(name="codexGroup", force=True)

print('refresh ok')

e = client.train.setup_project_experiment(
    experiment_name='test',
    model_type='Classification',
    group='codexGroup',
)

print(e)

PATH = '<path_to_sample>/'

train_type = CustomTrain(
    training_reference='train_model',
    run_name='more test',
    python_version='3.10',
    input_data=PATH + 'dados.csv',
    source=PATH + 'app.py',
    requirements=PATH + 'requirements.txt',
)

print(train_type)

exe = client.train.run(
    experiment=e,
    train_type=train_type,
    wait_ready=True
)

print(exe)

print(exe.status)

[ceciliacsilva]

Is the codex only used outside of production in development? Just to confirm that the standard is being maintained.

[ceciliacsilva]

The idea is to refactor the rest of the modules to use this?