Apache Spark

At this moment relu_layer op doesn't allow threshold configuration, and legacy RELU op allows that.
We should add configuration option to relu_layer.

Describe the bug

Serverless: Deprecation warning: Variables resolver reports following resolution errors:
              - Cannot resolve variable at "provider.environment.CUBEJS_APP": Value not found at "self" source,
              - Cannot resolve variable at "functions.cubejsProcess.events.0.event.resource": Value not found at "self" source
            From a next major this will be

Hi,
I am running the deltaTable = DeltaTable.convertToDelta(spark, f"parquet.{data_path}") to read a DeltaTable from the parquet files but it doesn't return one as suggested in the documents. However, it successfully converts them. If I read them exactly after that line again using forPath, it will give me the DeltaTable.

![image](https://user-images.githubusercontent.com/3591686/12765840

Used Spark version
Spark Version: 2.4.4
Used Spark Job Server version
SJS version: v0.11.1

Deployed mode
client on Spark Standalone

Actual (wrong) behavior
I can't get config, when post a job with 'sync=true'. I got it：
http://localhost:8090/jobs/ff99479b-e59c-4215-b17d-4058f8d97d25/config
{"status":"ERROR","result":"No such job ID ff99479b-e59c-4215-b17d-4058f8d97d25"

I have a simple regression task (using a LightGBMRegressor) where I want to penalize negative predictions more than positive ones. Is there a way to achieve this with the default regression LightGBM objectives (see https://lightgbm.readthedocs.io/en/latest/Parameters.html)? If not, is it somehow possible to define (many example for default LightGBM model) and pass a custom regression objective?