[Milestone 4] Optimize Spark SQL performance

### Task Description

**What needs to be done:**
Analyze the query plan when deserializing the data and make sure that this happens after any filtering on the structured data columns and after any joins or other shuffle steps.

**Why this task is needed:**
This will help reduce the cost of jobs that deal with unstructured data.

### Task Type

Code improvement/refactoring

### Related Issues

**Parent feature issue:** (if applicable )
**Related issues:**
NOTE: Use `Relationships` button to add parent/blocking issues after issue is created.

Start by analyzing the Spark SQL query plan around data deserialization, filtering on structured columns, joins, and other shuffle steps. Done means deserialization is deferred until after those operations, reducing the cost of jobs handling unstructured data.

Written by the indexing model from the issue text.