[Bug] Using the switch foundry server start --port xxxx causes foundry model list to return less models

### Describe the issue

When you start foundry server with the switch --port xxxx or --idle-timeout xxxx causes a bug so when you execute foundry model list NPU models are not listed and some times not even GPU Models. If you start the server without either of those switches foundry model list returns all possible models, NPU GPU CPU

### To reproduce

When you start foundry server with the switch --port xxxx or --idle-timeout xxxx causes a bug so when you execute foundry model list NPU models are not listed and some times not even GPU Models. If you start the server without either of those switches foundry model list returns all possible models, NPU GPU CPU

### Urgency

_No response_

### Platform and architecture

Windows X64

### OS Version

Windows 11 25H2

### Installation type

Released package/binary

### Foundry Local version

0.10.3

### API or surface area

CLI

### Hardware acceleration/backend

WinML EPs

### Backend/runtime version

NA

Reproduce the issue through the CLI with `foundry server start --port xxxx` and `--idle-timeout xxxx`, then compare `foundry model list` with a server started without those switches. Trace the server-start and model-list entry points; done means NPU, GPU, and CPU models are listed consistently in all cases.

Written by the indexing model from the issue text.