[FEATURE]: ensure Spark.sql-like code is not migrated accidentally

[ericvergnaud]

Is there an existing issue for this?

• I have searched the existing issues

Problem statement

Our current implementation of code migration assumes that a Spark.sql method call is eligible for migration based on its names and a few other checks: number of arguments, type of arguments... It does not check whether the method genuinely belongs to the Spark.sql API.
This may very well be good enough, but if user code contains code that closely mimics some Spark.sql API calls, then they could be migrated accidentally.

Proposed Solution

Implement context analysis in SparkSql class

Additional Context

No response