Refactor the DeepSpeed strategy config management

Nobody has claimed this yet.

### Outline & Motivation

DeepSpeed works by using a configuration file (dictionary) that allows customizing all of its aspects: https://www.deepspeed.ai/docs/config-json/

The `DeepSpeedStrategy` supports two ways of defining this:

1. Passing a config file, where every other argument becomes unused: https://github.com/Lightning-AI/lightning/blob/b792c90ea7148d61af192fde6c338ebbd355702f/src/lightning/fabric/strategies/deepspeed.py#L191
2. Exposes multiple of these arguments in the `__init__` that are used to define a base config. https://github.com/Lightning-AI/lightning/blob/b792c90ea7148d61af192fde6c338ebbd355702f/src/lightning/fabric/strategies/deepspeed.py#L242-L271

Option 2 is not scalable because:
- It forces us to duplicate all arguments
- Our docstrings might become outdated
- Our strategy defaults might diverge from the defaults in deepspeed
- It forces the user to either create an entire config or use these arguments
- Arguments might be different based on the installed deepspeed version as we support more than a single version.
- When deepspeed adds an argument that we don't expose, users have to switch to using the config

### Pitch

Remove all these exposed arguments and just have a `config` argument that overloads support for:
- Passing a path to a config file
```python
DeepSpeedStrategy(config="my/config/path.json")
```
- Passing a full config object
```python
config = ds.runtime.config.DeepSpeedConfig({"train_micro_batch_size_per_gpu": 2})
DeepSpeedStrategy(config=config)
```
- Passing a config dictionary (or a subset of it) that will update the default config
```python
config = {"zero_optimization": {"offload_optimizer": {"device": "cpu"}}}
DeepSpeedStrategy(config=config)
```

Where the default config is created by calling: https://github.com/microsoft/DeepSpeed/blob/085981bf1caf5d7d0b26d05f7c7e9487e1b35190/deepspeed/runtime/config.py#L674

### Additional context

DeepSpeed is considered experimental so we could do this breaking change: https://github.com/Lightning-AI/lightning/blob/b792c90ea7148d61af192fde6c338ebbd355702f/src/lightning/fabric/strategies/deepspeed.py#L99

cc @justusschock @awaelchli @carmocca

Start in src/lightning/fabric/strategies/deepspeed.py, especially the DeepSpeedStrategy initializer and the existing config handling around the referenced lines. Review DeepSpeed's runtime/config.py defaults and determine how path, full config object, and partial dictionary inputs should be supported. Done means the exposed configuration arguments are removed and all three config forms update or use the default configuration as described.

Written by the indexing model from the issue text.