Support for TDM

## Motivation

Add TDM (Tensor Data Movement) programming support for gfx1250 and fix
compilation issues preventing TDM tests from running.

## Technical Details

- Added amd_gfx1250_TDM.h header with device-specific TDM structures
- Implemented basic TDM load/store unit test for 2D data movement
- Fixed preprocessor directive in test - removed __gfx1250__ check that
was causing tests to be excluded during host compilation

## JIRA ID

SWDEV-575524

## Test Plan

Basic TDM load/store test validates 2D data movement between global
memory and shared memory.

## Test Result

Basic TDM load/store test validates 2D data movement between global
memory and shared memory on gfx1250.

## Submission Checklist

- [ X] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
[rocm-systems] ROCm/rocm-systems#6656 (commit bd33791)