Metric AI service

[tybalex]

This is the python service implementation of the Epic attached.
a few concepts in this service:

• A job: A list of namespaces of a cluster to be watched, selected by user, similar to the watchlist we already have in log anomaly detection
• A job run: run metric analysis for a job at the requested timestamp.

What's included in this PR:

1. Http server in main.py: provide a few APIs, the most important one is /run_job/{job_id}/, which is the logic to run a job. The other APIs are used by gateway plugin.
2. filter_anomaly_metric.py defines the rules to filter out normal metrics.
3. the model dir, cnn_model.py defines the structure of the CNN model and metric_pattern_classification.py defines the logic to train/eval/predict with the model.
4. Grafana_dashboard_utils.py defines functions to create/delete dashboardjson.
5. the tests dir: unit-test

[tybalex]

Regard to the model file, @AmartC will move it to Opni's internal S3 bucket.