feat: update static models to use 0.7.11 models

[sthaha]

NOTE: intel_rapl_DynPower has not been updated since
SGDRegressor for Dynamic Power Node Type 0 is missing https://github.com/sustainable-computing-io/kepler-model-db/tree/main/models/v0.7/ec2-0.7.11/rapl-sysfs/DynPower/BPFOnly

After updating the models to the latest, here are the changes on my machine

kepler_node_info{components_power_source="rapl-sysfs", cpu_architecture="Skylake", instance="kepler-latest:8888", job="latest", platform_power_source="none", source="os"} | 1
-- | --
kepler_node_info{components_power_source="estimator", cpu_architecture="Skylake", instance="kepler-dev:8888", job="dev", platform_power_source="none", source="os"}

NOTE: kepler-latest reports power_sources incorrectly (fixed in this pr; see "dev")

Platform / Idle

• seems like an improvement

Platform / Dynamic

Package / Idle

• Idle power is much higher than the old

Package / dynamic

• higher but closer to real usage

[github-actions]

🤖 SeineSailor

Here is a concise summary of the pull request changes:

Summary: This pull request updates static models to use version 0.7.11, but it's still in draft stage due to issues with model loading. The changes primarily affect the power estimation functionality, with key modifications including:

1. Updated power model initialization: The createNodeComponentPowerModelConfig function now accepts three string slices as arguments, and the CreateNodeComponentPoweEstimatorModel function initializes the nodeComponentPowerModel variable using the createPowerModelEstimator function.
2. Error handling and logging: The CreateNodePlatformPoweEstimatorModel function now checks for errors when creating the power model estimator and logs detailed error messages. Logging has been improved to provide more informative messages.
3. New ModelOutputType and SourceURL function: A new type for ModelOutputType has been added, and a new SourceURL function is added to the ModelConfig struct, which returns the URL or filepath of the init model based on whether InitModelURL or InitModelFilepath is set.
4. Simplified model weight copying: The prepare_dev_env.sh script now copies all .json files from the ../data/model_weight directory to /var/lib/kepler/data/model_weight, simplifying the process and ensuring inclusion of all necessary models.
5. Config changes: The GetModelConfigMap() function now trims string values to remove leading/trailing whitespaces, and a new GetDefaultPowerModel function returns the trainer and the path to the embedded power model based on modelOutputType and energySource.
6. Expose estimated idle power metrics: The kepler.config files now set EXPOSE_ESTIMATED_IDLE_POWER_METRICS to true.

Observations and suggestions:

• The changes seem to focus on updating the power estimation functionality, but the pull request is still in draft stage due to issues with model loading. It's essential to resolve these issues before merging the changes.
• The addition of the GetDefaultPowerModel function and the SourceURL function to the ModelConfig struct provides more flexibility and configurability to the power estimation functionality.
• The changes to the prepare_dev_env.sh script simplify the model weight copying process, but it's crucial to ensure that all necessary models are included and correctly loaded.
• The logging improvements will provide more informative messages, which can aid in debugging and troubleshooting.
• It's recommended to thoroughly test the updated power estimation functionality to ensure it works correctly and doesn't introduce any regressions.

[sunya-ch]

/lgtm Thank you so much!