feat(version-tests): Add comprehensive API coverage to TagsTester

[Liljis]

Description

Expanded the version compatibility tests in TagsTester.java to include comprehensive API coverage beyond just the convert source endpoint, as requested in #134.

Changes Made

Added 5 new test methods to doConversion() in TagsTester.java:

1. testConvertWithMultipleFormats - Tests conversion with MARKDOWN, JSON, and TEXT output formats ( already existed, done only refactoring)

2. testConvertWithJsonFormat - Validates JSON output and DoclingDocument deserialization

• Requests only JSON format
• Tests: "Can we properly deserialize the JSON into a DoclingDocument Java object?"
• Validates: The JSON structure is correct and can be parsed into the Java class
• Checks: Specific fields like getName() work correctly

3. testConvertWithDifferentOptions - Tests conversion with OCR, images, table mode, and custom timeout

• Extracts text from images in the document
• Embeds or references images in the converted output
• Controls how tables are extracted and formatted
• Sets maximum time for document processing

4. testClearConverters - Tests the /v1/clear/converters endpoint

• Clears cached converter instances (PDF parsers, OCR engines, etc.)

5. testClearResults - Tests the /v1/clear/results endpoint

• Clears cached conversion results (the actual converted documents)

Motivation

The previous implementation only tested the basic convert source API. This enhancement provides:

• Better coverage of the Docling Serve API surface
• Validation of multiple endpoints (health, convert, clear)
• Testing of various conversion options and output formats
• More comprehensive compatibility checks across versions

Test Results

• Build successful with no compilation errors
• All v1.x versions (v1.0.0+) pass all 6 tests
• v0.x versions fail as expected (AP
  results.md