EMA validation swap models before first update_parameters

Nobody has claimed this yet.

### Bug description

With `EMAWeightAveraging` and a delayed start (e.g. update_starting_at_step: 1000), validation still swaps `pl_module` and `_average_model` on every val epoch (e.g. val every 100 steps), even while `self._average_model.n_averaged == 0`. The `_average_model` is only a deepcopy(pl_module) from setup() and has never been updated. Retrieval metrics before `update_starting_at_step` near 0.

Example for EMA init:
```yaml:
use_buffers: false
decay: 0.999
update_every_n_steps: 1
update_starting_at_step: 1000
update_starting_at_epoch: -1
```

Suggested fix: in `WeightAveraging` [on_validation_epoch_start](https://github.com/Lightning-AI/pytorch-lightning/blob/master/src/lightning/pytorch/callbacks/weight_averaging.py#L216) / [on_validation_epoch_end](https://github.com/Lightning-AI/pytorch-lightning/blob/master/src/lightning/pytorch/callbacks/weight_averaging.py#L230), only call `_swap_models` when `self._average_model.n_averaged > 0`.

### What version are you seeing the problem on?

v2.6

### Reproduced in studio

_No response_

### How to reproduce the bug

```python

```

### Error messages and logs

_No response_

### Environment

_No response_

### More info

_No response_

cc @ethanwharris

Start in src/lightning/pytorch/callbacks/weight_averaging.py at on_validation_epoch_start and on_validation_epoch_end, then inspect how _average_model.n_averaged is initialized and updated. Reproduce the delayed-start configuration from the issue and verify validation does not swap the models before averaging begins, while swapping still occurs after n_averaged becomes positive.

Written by the indexing model from the issue text.