Add maxcompute batch predictor

[shydefoo]

Description

• This PR adds MaxCompute as a source and sink to batch prediction jobs. This allows users to perform batch prediction on data stored in MaxCompute and write the results back into MaxCompute.

Modifications

• python/batch-predictor
  • merlinpyspark package source code modified
  • Added missing proto spec for PredictionJob protobuf
  • Added MaxCompute Jar dependencies to base.Dockerfile
• python/sdk
  • Modified python sdk to include MaxComputeSource and MaxComputeSink classes
  • Modified swagger.yaml and regenerated python and go clients
  • Will add tests and readme in separate MR

Tests

Checklist

• Added PR label
• Added unit test, integration, and/or e2e tests
• Tested locally
• Updated documentation
• Update Swagger spec if the PR introduce API changes
• Regenerated Golang and Python client if the PR introduces API changes

Release Notes

- Support MaxCompute source and sink for batch prediction jobs

[shydefoo]

@deadlycoconuts @bthari poke

[deadlycoconuts]

Thanks a lot for all the changes! 🚀 Especially the reverse engineering of the proto and adding the additional check on the protoc version. Everything looks good, I just left random nitpicking comments, but thanks a lot!