Introduce isolated integration testing pipeline to reduce manual release validation effort

Now everything is in the https://github.com/zaleslaw/KDA-Test-Projects

### Problem

Currently, validation of DataFrame changes before releases relies heavily on manual testing using external projects (e.g. KDA-Test-Projects).

This leads to:
- 5–7 person-hours of manual validation per mini-release
- high cost of frequent releases
- risk of missing regressions across real-world usage scenarios

### Expected

Introduce an isolated integration testing pipeline that automatically validates DataFrame against external projects and realistic usage scenarios.

### Acceptance criteria

- External test projects are integrated into a dedicated CI pipeline
- Pipeline runs automatically for release candidates / key branches
- Tests cover real-world usage (not only unit tests)
- Failures are visible and block release if necessary
- Manual validation effort is significantly reduced

### Motivation

- Multiple releases are expected before and after 1.0
- Current manual validation does not scale
- Integration testing ensures API stability across real use cases
- Reduces release cost by 3–5x (from ~5–7h to ~1–2h)
- Increases confidence that nothing is broken before shipping