Reject invalid bucket and truncate transform parameters

### Problem

`Transform::from_str` currently accepts invalid parameterized transform strings. For example, `bucket[0]` and `truncate[0]` deserialize successfully even though applying those transforms can later panic due to modulo or division by zero. The parser also accepts malformed strings such as `bucket10` and `truncate10` because it strips the transform prefix and trims brackets loosely.

### Expected behavior

Parameterized transform parsing should require exact Iceberg syntax and validate the parameter value:

- accept `bucket[N]` and `truncate[W]` when the value is greater than zero
- reject zero values such as `bucket[0]` and `truncate[0]`
- reject malformed syntax such as `bucket10`, `truncate10`, or trailing characters after the closing bracket

### Impact

Invalid table metadata or user-provided transform strings can be accepted and fail later during transform evaluation. Rejecting them during parsing gives callers a normal validation error instead of allowing a later runtime panic.

Start at Transform::from_str and inspect the existing parameterized-transform parsing and its tests. Verify that bucket[N] and truncate[W] accept only positive values and that malformed or trailing input is rejected. Done means invalid strings return a normal validation error instead of being accepted.

Written by the indexing model from the issue text.