[cli] indexing: improve dataset selectors

### StatGPT Backend version

0.3.0

### What is the problem this feature will solve?

- select multiple datasets to index at once (multiselect mode)
- index all "NOT_STARTED" datasets at once, without the need to manually select them

### What is the proposed feature or solution?

can:
- add new if/else blocks
- or refactor selectors in a class and support filtering by dataset attributes (like: `status=NOT_STARTED`, `older_than="2026-01-01"`)

### What alternatives have you considered?

_No response_

The issue names CLI dataset indexing and selectors but no files or tests. Start by locating the current selector entry point and its tests, then trace how datasets are selected for indexing. Done means supporting multiple selected datasets and indexing all NOT_STARTED datasets, with the final filtering scope clarified before implementation.

Written by the indexing model from the issue text.