"save_last" could not save a complete checkpoint

Nobody has claimed this yet.

### Bug description

I use save_last=True and every_n_epoch = 1 ( actually the default setting).

every_n_epoch = 1 can lead to correct checkpoint saving, everything is ok.

But the last.ckpt is incomplete, the epoch is 0 and the global is 0.

The file size of last.ckpt is around 800M while epoch*.ckpt is around 2.4g.

My pl version is 1.9.3 and torch version is 2.2.1

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

v1.9

### How to reproduce the bug

_No response_

### 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_

Start by reproducing the report on PyTorch Lightning v1.9.3 with save_last=True and every_n_epoch=1, then compare last.ckpt with the epoch*.ckpt files. Trace the checkpoint-saving path to determine why last.ckpt contains epoch and global step as 0; done means last.ckpt is complete and matches the saved checkpoint state.

Written by the indexing model from the issue text.