FloatingPointValue unit of least precision (ulp) is incorrect for subnormals

### Describe the bug

the ulp function is incomplete and does not properly compute the unit of least precision when the number
is subnormal.

### To Reproduce

_No response_

### Expected behavior

_No response_

### Actual behavior

_No response_

### Additional: Dart SDK info

_No response_

### Additional: pubspec.yaml

```yaml

```

### Additional: Context

_No response_

The issue names FloatingPointValue and its ulp function but no files or tests. Start by locating that implementation and its existing numeric tests, then add coverage for subnormal values and verify that the computed unit of least precision matches the expected floating-point behavior.

Written by the indexing model from the issue text.