Describe the issue:
During computing Channel Dependencies reshape_break_channel_dependency does following code to ensure that the number of input channels equals the number of output channels:

in_shape = op_node.auxiliary['in_shape']
out_shape = op_node.auxiliary['out_shape']
in_channel = in_shape[1]
out_channel = out_shape[1]
return in_channel != out_channel

This is correct

Once Woodwork implements this issue, we can clean up the Woodwork initialization in add_last_time_indexes to pass in the previous dataframe's table schema to keep that typing information but also perform inference on the new last time index column.

Is your feature request related to a problem? Please describe.
Currently in feature_store.yaml, we can only specify a region for DynamoDB provider. As a result, it requires an actual DynamoDB to be available when we want to do local development/testing or integration testing in a sandbox environment.

Describe the solution you'd like
A way to solve this is to let user pass an endpoint

Problem
Some of our transformers & estimators are not thoroughly tested or not tested at all.

Solution
Use OpTransformerSpec and OpEstimatorSpec base test specs to provide tests for all existing transformers & estimators.

When using r2 as eval metric for regression task (with 'Explain' mode) the metric values reported in Leaderboard (at README.md file) are multiplied by -1.
For instance, the metric value for some model shown in the Leaderboard is -0.41, while when clicking the model name leads to the detailed results page - and there the value of r2 is 0.41.
I've noticed that when one of R2 metric values in the L

https://github.com/4paradigm/OpenMLDB/blob/420398b8349880bce262db06c3b143d023b56862/java/openmldb-batch/src/main/scala/com/_4paradigm/openmldb/batch/nodes/SelectIntoPlan.scala#L39-L41

If select closure in select into result is empty, we just skip saving, and the job is succeed. But users can't find the output file. It's not intuitive.
We should throw an exception.

The transformer should create computations over windows of past values of the features, and populate them at time t, t being the time of the forecast.

It uses pandas rolling, outputs several comptutations, mean, max, std, etc, and pandas shift to move the computations to the right row.

tmp = (data[variables]
       .rolling(window='3H').mean()  # Average the last 3 hr values.
       .

In #3324 , we had to mark some tests as expected to fail since XGBoost was throwing a FutureWarning. The warning has been addressed in XGBoost, so we're just waiting for the PR merged to be released. This issue is discussed in the #3275 issue.

evalml/tests/component_tests/test_xgboost_classifier.py needs to have the @pytest.mark.xfail removed f

what

Ray workflows seems like something we could easily add too https://docs.ray.io/en/latest/workflows/concepts.html given that we now have GraphAdapters

Task

• Implement something very similar to the RayGraphAdapter, i.e. RayWorkflowGraphAdapter. The hypothesis is that then we just need to use workflow step function to wrap hamilton functions.
• implement an integration test f