Support special test parametrizations

[carmocca]

What does this PR do?

Second attempt after #10462

I have rolled back some of the parameterizations. If you know of any others, feel free to push a change to this PR.

The current list of tests is:

================================================================================
Ran	tests/accelerators/test_accelerator_connector.py::test_accelerator_choice_ddp_cpu_and_strategy
Ran	tests/accelerators/test_ddp.py::test_ddp_wrapper[16]
Ran	tests/accelerators/test_ddp.py::test_ddp_wrapper[32]
Ran	tests/accelerators/test_multi_nodes_gpu.py::test_logging_sync_dist_true_ddp
Ran	tests/accelerators/test_multi_nodes_gpu.py::test__validation_step__log
Ran	tests/callbacks/test_pruning.py::test_pruning_callback_ddp[False-False]
Ran	tests/callbacks/test_pruning.py::test_pruning_callback_ddp[False-True]
Ran	tests/callbacks/test_pruning.py::test_pruning_callback_ddp[True-False]
Ran	tests/callbacks/test_pruning.py::test_pruning_callback_ddp[True-True]
Ran	tests/callbacks/test_stochastic_weight_avg.py::test_swa_callback_ddp
Ran	tests/callbacks/test_tqdm_progress_bar.py::test_progress_bar_max_val_check_interval[8-4-2-1-0.2]
Ran	tests/callbacks/test_tqdm_progress_bar.py::test_progress_bar_max_val_check_interval[8-4-2-1-0.5]
Ran	tests/checkpointing/test_checkpoint_callback_frequency.py::test_top_k_ddp[1-1-1.0-1]
Ran	tests/checkpointing/test_checkpoint_callback_frequency.py::test_top_k_ddp[2-2-0.3-4]
Ran	tests/core/test_metric_result_integration.py::test_result_collection_reload_2_gpus
Ran	tests/lite/test_lite.py::test_deepspeed_multiple_models
Ran	tests/lite/test_parity.py::test_boring_lite_model_ddp[32-ddp-2-gpu]
Ran	tests/models/test_hooks.py::test_transfer_batch_hook_ddp
Ran	tests/models/test_hooks.py::test_trainer_model_hook_system_fit_deepspeed[True]
Ran	tests/models/test_hooks.py::test_trainer_model_hook_system_fit_deepspeed[False]
Ran	tests/models/test_sync_batchnorm.py::test_sync_batchnorm_ddp
Ran	tests/plugins/test_amp_plugins.py::test_amp_apex_ddp_fit[O2]
Ran	tests/plugins/test_ddp_fully_sharded_with_full_state_dict.py::test_fully_sharded_plugin_checkpoint
Ran	tests/plugins/test_ddp_fully_sharded_with_full_state_dict.py::test_fully_sharded_plugin_checkpoint_multi_gpus
Ran	tests/plugins/test_ddp_fully_sharded_with_full_state_dict.py::test_fsdp_gradient_clipping_raises
Ran	tests/plugins/test_ddp_plugin.py::test_ddp_with_2_gpus
Ran	tests/plugins/test_ddp_plugin.py::test_ddp_barrier_non_consecutive_device_ids
Ran	tests/plugins/test_ddp_plugin_with_comm_hook.py::test_ddp_fp16_compress_comm_hook
Ran	tests/plugins/test_ddp_plugin_with_comm_hook.py::test_ddp_sgd_comm_hook
Ran	tests/plugins/test_ddp_plugin_with_comm_hook.py::test_ddp_fp16_compress_wrap_sgd_comm_hook
Ran	tests/plugins/test_ddp_plugin_with_comm_hook.py::test_ddp_spawn_fp16_compress_comm_hook
Ran	tests/plugins/test_ddp_plugin_with_comm_hook.py::test_ddp_post_local_sgd_comm_hook
Ran	tests/plugins/test_deepspeed_plugin.py::test_warn_deepspeed_ignored
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_run_configure_optimizers
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_config
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_custom_precision_params
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_fp32_works
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_stage_3_save_warning
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_single_file
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_stage_3
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_stage_3_manual_optimization
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_stage_3_checkpointing
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_stage_3_warns_resume_training
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_stage_3_resume_training
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_stage_3_checkpointing_full_weights_manual
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_stage_2_accumulated_grad_batches
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_stage_2_accumulated_grad_batches_offload_optimizer
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_test
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_partial_partition_parameters
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_test_rnn
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_multigpu_no_schedulers
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_skip_backward_raises
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_setup_train_dataloader
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_scheduler_step_count
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_scheduler_step_count_epoch
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_configure_gradient_clipping
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_gradient_clip_by_value
Ran	tests/plugins/test_deepspeed_plugin.py::test_different_accumulate_grad_batches_fails
Ran	tests/plugins/test_deepspeed_plugin.py::test_specific_gpu_device_id
Ran	tests/plugins/test_deepspeed_plugin.py::test_deepspeed_with_meta_device
Ran	tests/plugins/test_sharded_plugin.py::test_ddp_sharded_plugin_test_multigpu[trainer_kwargs0]
Ran	tests/plugins/test_sharded_plugin.py::test_ddp_sharded_plugin_test_multigpu[trainer_kwargs1]
Ran	tests/plugins/test_sharded_plugin.py::test_ddp_sharded_plugin_manual_optimization_spawn
Ran	tests/plugins/test_sharded_plugin.py::test_ddp_sharded_plugin_manual_optimization
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_ddp
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[ManualOptimBoringModel-1]
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[ManualOptimBoringModel-2]
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[ManualOptimBoringModel-3]
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[ManualOptimBoringModel-4]
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[ManualOptimBoringModel-5]
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[BoringModel-1]
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[BoringModel-2]
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[BoringModel-3]
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[BoringModel-4]
Ran	tests/profiler/test_profiler.py::test_pytorch_profiler_trainer_fit[BoringModel-5]
Skipped	tests/profiler/test_profiler.py::test_pytorch_profiler_nested_emit_nvtx
Ran	tests/trainer/logging_/test_train_loop_logging.py::test_logging_sync_dist_true_ddp
Ran	tests/trainer/optimization/test_manual_optimization.py::test_step_with_optimizer_closure_with_different_frequencies_ddp
Ran	tests/trainer/optimization/test_manual_optimization.py::test_step_with_optimizer_closure_with_different_frequencies_ddp_with_toggle_model
Ran	tests/trainer/optimization/test_optimizers.py::test_optimizer_state_on_device
Ran	tests/trainer/test_trainer.py::test_trainer_predict_special[kwargs0]
Ran	tests/trainer/test_trainer.py::test_trainer_predict_special[kwargs1]
Ran	tests/trainer/test_trainer.py::test_trainer_predict_special[kwargs2]
Ran	tests/trainer/test_trainer.py::test_ddp_terminate_when_deadlock_is_detected
Ran	tests/utilities/test_all_gather_grad.py::test_all_gather_collection
Ran	tests/utilities/test_all_gather_grad.py::test_all_gather_sync_grads
Ran	tests/utilities/test_deepspeed_collate_checkpoint.py::test_deepspeed_collate_checkpoint
Ran	tests/utilities/test_meta.py::test_init_meta_context
Ran	tests/utilities/test_warnings.py
Ran	manual ddp launch test
================================================================================

Does your PR introduce any breaking changes? If yes, please list them.

None

Before submitting

• Was this discussed/approved via a GitHub issue? (not for typos and docs)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• [n/a] Did you make sure to update the documentation with your changes? (if necessary)
• Did you write any new necessary tests? (not for typos and docs)
• Did you verify new and existing tests pass locally with your changes?
• Did you list all the breaking changes introduced by this pull request?
• [n/a] Did you update the CHANGELOG? (not for typos, docs, test updates, or internal minor changes/refactorings)

PR review

• Is this pull request ready for review? (if not, please submit in draft mode)
• Check that all items from Before submitting are resolved
• Make sure the title is self-explanatory and the description concisely explains the PR
• Add labels and milestones (and optionally projects) to the PR so it can be classified

cc @carmocca @akihironitta @Borda