MLFlow Logger log_model parameter not working correctly when using multiple ModelCheckpoint

Nobody has claimed this yet.

### Bug description

MLFlow Logger parameter 'log_model' is not working as expected when multiple ModelCheckpoint are used.
When set to 'True' we expect it to save one checkpoint per ModelCheckpoint used at the end of the training. But what we get is only one of the model checkpoints is logged.
****

### How to reproduce the bug

```python
- Define pl.Trainer with 2 checkpoint callbacks (1 "LastCheckpoint", without save_top k + 1 BestCheckpoint, with save_top k=n>1)
- Define MLFlowLogger with log_model set to True
- get only 1 checkpoint as a result in MLFlow

This is probably due to after_save_checkpoint() from MLFlowLogger that overwrite checkpoint_callback everytime, losing the one of the two ModelCheckpoint reference. It may be necessary to use a list here.
```

### Error messages and logs

```
# Error messages and logs here please
```

### Environment

Current environment

```
#- Lightning Component (e.g. Trainer, LightningModule, LightningApp, LightningWork, LightningFlow):
#- PyTorch Lightning Version (e.g., 1.5.0):
#- Lightning App Version (e.g., 0.5.2):
#- PyTorch Version (e.g., 2.0):
#- Python version (e.g., 3.9):
#- OS (e.g., Linux):
#- CUDA/cuDNN version:
#- GPU models and configuration:
#- How you installed Lightning(`conda`, `pip`, source):
#- Running environment of LightningApp (e.g. local, cloud):
```

### More info

_No response_

Reproduce the issue with two ModelCheckpoint callbacks and MLFlowLogger(log_model=True), then inspect MLFlowLogger.after_save_checkpoint() and how it tracks checkpoint callbacks. Confirm the fix by verifying that one checkpoint from each ModelCheckpoint is logged in MLflow.

Written by the indexing model from the issue text.