Streaming Ingestion Pipeline with Spark

[pyalex]

What this PR does / why we need it:

This PR replaces current (beam) Ingestion Job with spark version. However, several changes are introduced:

1. Input to streaming pipeline is arbitrary* protobuf. Message class must be provided through FeatureTable configuration.
2. Ingestion job stores only to redis and writes deadletters as parquet files.

Currently there are two ways to make protobuf class available for job in runtime:

1. Compile proto & pack it into jar and add jar to list of files on Spark-submit.
   This option has a limitation that compiled class must be linked to shadowed protobuf library com.google.protobuf.vendor
   since in our job we had to shadow protobuf due to conflicts with spark dependencies.
2. Use proto registry (external service) as source of proto descriptors. As starter we gonna support https://github.com/gojekfarm/stencil But we would need to support more common implementation of proto registry.

*Arbitrary protobuf has some limitations. oneof, google.protobuf.Any are not supported. Thus FeatureRow can't be input to new streaming pipeline.

Which issue(s) this PR fixes:

Fixes #

Does this PR introduce a user-facing change?:

Streaming ingestion job expects arbitrary protobuf as input with some limitations to not supporting `oneof` since spark's dataframe doesn't support it

[feast-ci-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: oavdeev, pyalex

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [pyalex]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[woop]

/lgtm