Make AbstractAgentSet.select() and CellCollection.select() raise on negative integer or boolean at_most

### Describe the bug
Following PR #3853 and PR #3860 which added float fraction validation in `select()`, negative integers and boolean inputs are still not validated in `AbstractAgentSet.select()` and `CellCollection.select()`.

### To Reproduce
```python
# 1. Negative integer silently returns empty set instead of raising ValueError
model.agents.select(at_most=-1) # returns 0 agents

# 2. Booleans are silently treated as 0 or 1 because isinstance(bool, int) is True
model.agents.select(at_most=False) # returns 0 agents
model.agents.select(at_most=True) # returns 1 agent

# 3. Non-numeric types crash late inside generator rather than failing eagerly
list(model.agents.select(at_most="5")) # TypeError: '>=' not supported between instances of 'int' and 'str'
```

*(The same behavior reproduces on `CellCollection.select()`)*.

### Expected Behavior
- Passing an integer `< 0` should raise `ValueError("Integer at_most must be non-negative, got ...")`.
- Passing a boolean or non-number should raise `TypeError("at_most must be an integer or float, got ...")`, matching `AbstractAgentSet.select_random()`.

I have a fix with unit tests ready for both modules and can open a PR.

Start with AbstractAgentSet.select() and CellCollection.select(), comparing their validation behavior with select_random(). Add unit tests for negative integers, booleans, and non-numeric values, then verify the methods raise the specified ValueError or TypeError before iteration.

Written by the indexing model from the issue text.