feat(bigquery): allow the user to ask for skipping parsing rows when quering

[rafal-hawrylak]

Summary
This PR introduces a new optional boolean flag, skipParsing, to the BigQuery.query, Table.getRows, and Job.getQueryResults methods.

When skipParsing is set to true, the library bypasses the internal mergeSchemaWithRows_ logic. Instead of returning idiomatic JavaScript objects (where BigQuery types are mapped to JS types), the rows result will contain the raw BigQuery API wire format (e.g., an array of { f: [{ v: '...' }] } objects).

Motivation
Providing access to raw rows offers two primary benefits:

1. Performance: For very large result sets, the overhead of merging the schema with rows and performing type conversions can be significant. Bypassing this step allows for faster processing when the consumer handles their own parsing.
2. Flexibility: Users who require the specific BigQuery JSON structure for downstream systems or custom serialization can now access it directly without the library's opinionated mapping.

[feywind]

/gcbrun