feat(import): add MLFlow import support (kit import mlflow://)

[akagami-harsh]

Description

Adds MLFlow as an import source for kit import, us to package MLFlow experiment runs as ModelKits in a single command. Supports three URI forms: mlflow://host/experiments/{id}/runs/{id}, mlflow://experiments/{id}/runs/{id}, and the short mlflow://runs/{id}. The tracking server is resolved from the URI host, the MLFLOW_TRACKING_URI env var, or defaults to http://localhost:5000. Authentication is handled via --token flag or MLFLOW_TRACKING_TOKEN env var. Only FINISHED runs can be imported RUNNING, FAILED, and KILLED runs are rejected with a clear error. Artifacts are downloaded concurrently through the tracking server's artifact proxy, a Kitfile is auto-generated (or a user-provided one is used), and MLFlow provenance metadata (run ID, experiment ID, metrics, params, timestamps) is injected into model.parameters before packing.

Resolves #1014

Linked issues

• Implement MLFlow import: kit import mlflow:// for experiment-to-ModelKit workflow #1014

AI-Assisted Code

• This PR contains AI-generated code that I have reviewed and tested
• I take full responsibility for all code in this PR, regardless of how it was created

[akagami-harsh]

tested on local setup:

• Start MLFlow locally

(venv) harshvir@msi:~/kitops$ mlflow server --host 0.0.0.0 --port 5000 
Backend store URI not provided. Using sqlite:///mlflow.db
Registry store URI not provided. Using backend store URI.
[MLflow] Security middleware enabled with default settings (localhost-only). To allow connections from other hosts, use --host 0.0.0.0 and configure --allowed-hosts and --cors-allowed-origins.
INFO:     Uvicorn running on http://0.0.0.0:5000 (Press CTRL+C to quit)
INFO:     Started parent process [82500]
2026/03/17 15:41:36 INFO mlflow.server.jobs.utils: Starting huey consumer for job function invoke_scorer
2026/03/17 15:41:36 INFO mlflow.server.jobs.utils: Starting huey consumer for job function run_online_trace_scorer

• Log a quick test run

(venv) harshvir@msi:~/kitops$ python3 testrun.py 
Run ID: 2de0dbba9b164eee9a6a27551afb6502
🏃 View run test-model at: http://localhost:5000/#/experiments/0/runs/2de0dbba9b164eee9a6a27551afb6502
🧪 View experiment at: http://localhost:5000/#/experiments/0

• Import the run using above run id

(venv) harshvir@msi:~/kitops$ ./kit import mlflow://runs/2de0dbba9b164eee9a6a27551afb6502 -t my-test-raw
:v1
Connecting to MLFlow tracking server http://localhost:5000
Found run "test-model" (experiment 0) — status FINISHED
Found 1 artifact file(s) for run 2de0dbba9b164eee9a6a27551afb6502
Downloading artifact model.pkl
Generated Kitfile:

manifestVersion: 1.0.0
package:
  name: test-model
  authors: [mlflow-exp0]
model:
  name: model
  path: model.pkl

Would you like to edit Kitfile before packing? (y/N): 
Updated Kitfile with MLFlow provenance:

manifestVersion: 1.0.0
package:
  name: test-model
  version: "20260317082144"
  description: Imported from MLFlow run 2de0dbba9b164eee9a6a27551afb6502 — (test-model) — tracking server http://localhost:5000
  authors: [mlflow-exp0]
model:
  name: model
  path: model.pkl
  parameters:
    mlflow_provenance:
      mlflow_end_time: "2026-03-17T08:21:44Z"
      mlflow_experiment_id: "0"
      mlflow_metrics:
        accuracy: 0.95
      mlflow_params:
        learning_rate: "0.01"
      mlflow_run_id: 2de0dbba9b164eee9a6a27551afb6502
      mlflow_run_name: test-model
      mlflow_run_status: FINISHED
      mlflow_start_time: "2026-03-17T08:21:44Z"
      mlflow_tracking_uri: http://localhost:5000

Packing model to my-test-raw:v1
Already saved model layer: sha256:81a20b25d26d08405a4cfb7e10862c400cb1792cff58d08b928c3a3428fbf77c
Saved configuration: sha256:d2437f172bf3eff84f8cbc29a14dd672e22cde50593bd139358f6be5f80943f3
Saved manifest to storage: sha256:771f1f22f20fcbdbe07a4604544d7d02bdebc72dd2f4f079aa80c19dfa8045df
Model is packed as my-test-raw:v1

• Check it was packed

(venv) harshvir@msi:~/kitops$ ./kit list
REPOSITORY      TAG      MAINTAINER    NAME         SIZE      DIGEST
model           v1       mlflow-exp0   test-model   2.0 KiB   sha256:9a8d2e31045803d283e382895d9689dce663071b27e8ac7c4d9803e45cd15b54
my-test-model   v5       mlflow-exp0   test-model   2.0 KiB   sha256:9a8d2e31045803d283e382895d9689dce663071b27e8ac7c4d9803e45cd15b54
my-test-model   <none>   mlflow-exp0   test-model   2.0 KiB   sha256:d72828c4ff7377e2eaf22ae030f3d5dd93a1b355ae87d9326f26d3f9deca1e48
my-test-model   v1       mlflow-exp0   test-model   2.0 KiB   sha256:e63c69003ab40166f64cf286610bfc57db1f83b71cdc18e845177073cacf18da
my-test-raw     v1       mlflow-exp0   test-model   2.0 KiB   sha256:771f1f22f20fcbdbe07a4604544d7d02bdebc72dd2f4f079aa80c19dfa8045df

(venv) harshvir@msi:~/kitops$ ./kit inspect my-test-raw:v1
{
  "digest": "sha256:771f1f22f20fcbdbe07a4604544d7d02bdebc72dd2f4f079aa80c19dfa8045df",
  "cliVersion": "unknown",
  "kitfile": {
    "manifestVersion": "1.0.0",
    "package": {
      "name": "test-model",
      "version": "20260317082144",
      "description": "Imported from MLFlow run 2de0dbba9b164eee9a6a27551afb6502 — (test-model) — tracking server http://localhost:5000",
      "authors": [
        "mlflow-exp0"
      ]
    },
    "model": {
      "name": "model",
      "path": "model.pkl",
      "parameters": {
        "mlflow_provenance": {
          "mlflow_end_time": "2026-03-17T08:21:44Z",
          "mlflow_experiment_id": "0",
          "mlflow_metrics": {
            "accuracy": 0.95
          },
          "mlflow_params": {
            "learning_rate": "0.01"
          },
          "mlflow_run_id": "2de0dbba9b164eee9a6a27551afb6502",
          "mlflow_run_name": "test-model",
          "mlflow_run_status": "FINISHED",
          "mlflow_start_time": "2026-03-17T08:21:44Z",
          "mlflow_tracking_uri": "http://localhost:5000"
        }
      },
      "digest": "sha256:81a20b25d26d08405a4cfb7e10862c400cb1792cff58d08b928c3a3428fbf77c",
      "diffId": "sha256:81a20b25d26d08405a4cfb7e10862c400cb1792cff58d08b928c3a3428fbf77c"
    }
  },
  "manifest": {
    "schemaVersion": 2,
    "artifactType": "application/vnd.kitops.modelkit.manifest.v1+json",
    "config": {
      "mediaType": "application/vnd.kitops.modelkit.config.v1+json",
      "digest": "sha256:d2437f172bf3eff84f8cbc29a14dd672e22cde50593bd139358f6be5f80943f3",
      "size": 838
    },
    "layers": [
      {
        "mediaType": "application/vnd.kitops.modelkit.model.v1.tar",
        "digest": "sha256:81a20b25d26d08405a4cfb7e10862c400cb1792cff58d08b928c3a3428fbf77c",
        "size": 2048,
        "annotations": {
          "org.cncf.model.filepath": "model.pkl"
        }
      }
    ],
    "annotations": {
      "ml.kitops.modelkit.cli-version": "unknown"
    }
  }
}

[akagami-harsh]

@gorkem, Could you please take a look again