PySpark DataFrame Column Analyzer

Static analyzer for PySpark DataFrame column usage in Python files.

Detects invalid column references for PySpark DataFrame operations:
- select()
- withColumn()
- withColumnRenamed()
- drop()
- join()
- union() / unionByName()
Infers DataFrame schemas from Palantir Foundry dataset metadata for the transform:
- spark.read.table("table_name") uses the schema derived from Foundry metadata
- The first select() on a DataFrame when no schema is known
Incremental analysis on document changes.
Nice-to-have features:
- Column completion inside df.select("...").
- Hover on a DataFrame variable name to see its inferred schema.

Schemas can come from:

a Palantir Foundry transform metadata file referenced by pysparkAnalyzer.foundryMetadataPath (extension-side preloaded map), or
live Foundry API calls from the Rust sidecar when FOUNDRY_TOKEN and FOUNDRY_API_URL is set.

For live sidecar fetches, set:

FOUNDRY_TOKEN (required): bearer token used for authorization.
FOUNDRY_API_URL (required): Foundry stack base URL, e.g. https://my-stack.palantirfoundry.com.
FOUNDRY_SCHEMA_CACHE_TTL_SECONDS (optional): in-memory cache TTL, default 300.

eredzik