Add a User Guide for the Parquet file reading

[zaleslaw]

Problem

Kotlin DataFrame already has a Parquet documentation page and a realistic example project:

• docs/StardustDocs/topics/dataSources/Parquet.md
• samples/src/test/kotlin/org/jetbrains/kotlinx/dataframe/samples/io/Parquet.kt
• examples/projects/spark-parquet-dataframe

However, the current docs are closer to an API reference for DataFrame.readParquet(...). We should add or expand a user-facing guide that explains how to read Parquet data in practical workflows, especially when Parquet files are produced by tools such as Apache Spark.

Goal

Create a user guide for reading Parquet files with Kotlin DataFrame, using the existing Spark + Parquet example as the main realistic scenario.

The guide should help users understand:

• when Parquet is useful;
• how to set up the required dependency/module;
• how to read one or multiple Parquet files with DataFrame.readParquet(...);
• how to read Spark-generated Parquet output, including part-*.parquet files;
• current limitations and platform constraints.

Suggested Scope

The guide should cover:

1. Setup

   • Mention that Parquet reading is provided through the dataframe-arrow module.
   • Clarify that it is included in the general dataframe artifact and in %use dataframe.

2. Basic reading examples

   • Read from a file path/string.
   • Read from Path.
   • Read from File.
   • Read from URL.
   • Read multiple Parquet files:

     val df = DataFrame.readParquet(file1, file2, file3)

3. Spark-produced Parquet

   • Explain that Spark commonly writes a Parquet dataset as a directory containing part-*.parquet files.
   • Show that Kotlin DataFrame currently reads concrete Parquet files, so users may need to pass the actual part files rather than the directory.
   • Reuse the existing example project:
     examples/projects/spark-parquet-dataframe.

4. Realistic workflow

   • Spark reads/prepares data.
   • Spark writes Parquet.
   • Kotlin DataFrame reads the exported Parquet files.
   • Kotlin DataFrame explores/transforms/prints the result.

5. Limitations

   • Kotlin DataFrame currently supports reading Parquet, not writing Parquet.
   • Android is not supported because Apache Arrow is not supported there.
   • Structured/nested Arrow types such as Struct are not supported yet.
   • Predicate pushdown / column pruning are not exposed through readParquet; users should select / filter after reading.

Acceptance Criteria

• The guide references or reuses validated snippets from samples/src/test/kotlin/org/jetbrains/kotlinx/dataframe/samples/io/Parquet.kt.
• The guide links to the existing Spark example project.
• The guide explains how to handle Spark output directories with part-*.parquet files.
• The limitations are clearly documented.
• The page is added to the docs navigation if a new guide page is created.
• Code snippets are validated through the existing docs sample mechanism where possible.