Enhance Interval arithmetic to support String type.

### Is your feature request related to a problem or challenge?

Currently DataFusion does not support Interval arithmetic on string type, and the stats `analyze` process will be failed during CBO stats estimation process.

I think maybe we can normalize the string intervals to numeric intervals first and leverage the numeric intervals to estimate the `selectivity`.

Init input Bound:
["aaa", "xyz123"] => [0, 1]

Estimated Bound (example)
["ccc", "ddd"] => [0.2, 0.3]

estimated selectivity:
0.1

The Normalization of the string intervals can be based on ascil code ordering, for any non-ascil char, we can just return early.

### Describe the solution you'd like

_No response_

### Describe alternatives you've considered

_No response_

### Additional context

_No response_

The issue points to the stats analyze process and CBO stats estimation; start by tracing the interval selectivity path and existing numeric interval handling there. Define how ASCII string bounds normalize, including the non-ASCII early-return case, then verify that analyze no longer fails and selectivity is estimated for string intervals.

Written by the indexing model from the issue text.