rv() rejects list time inputs despite documenting time as array_like

Nobody has claimed this yet.

## Summary

`ssapy.compute.rv()` documents `time` as `array_like`, but passing a normal Python list of GPS-second values fails before propagation.

## Reproduction

```python
import numpy as np
from ssapy import Orbit
from ssapy.compute import rv

orbit = Orbit(
np.array([7000e3, 0.0, 0.0]),
np.array([0.0, 7500.0, 0.0]),
0.0,
)

rv(orbit, [0.0, 60.0, 120.0])
```

`_countTime()` leaves an iterable input unchanged. `rv()` then wraps it in `HashableArrayContainer`, whose constructor accesses `self.arr.flags.writeable`. A Python `list` has no `.flags` attribute, so this raises `AttributeError`.

## Expected behaviour

List and tuple inputs should be accepted as documented `array_like` time inputs and normalised to an ndarray before they are used as cache keys.

## Proposed change

- Normalise non-`Time` iterable time inputs to NumPy arrays in `_countTime()`.
- Preserve the existing scalar-squeezing behaviour.
- Add regression coverage for list and tuple time inputs.

Start with `_countTime()` and follow how `rv()` passes time into `HashableArrayContainer`; the issue identifies the failing `.flags.writeable` access. Reproduce the list case, then check the existing scalar-squeezing behavior. Done means list and tuple GPS-second inputs work as documented and regression coverage verifies both forms.

Written by the indexing model from the issue text.