[Question]: Regarding Filtered Vector Search indexing and SSTable storage layout

I am evaluating SeekDB's performance in hybrid analytical and vector search scenarios. I have two specific questions regarding the underlying architecture:

1. Support for Hybrid/Filtered Vector Search Indexing:
   Does SeekDB currently support a unified hybrid index for filtered vector searches? Specifically, can the engine utilize both scalar filters and vector indexes simultaneously (e.g., using a single index structure or a bitmap-based pre-filter) to avoid the "pre-filtering vs. post-filtering" performance bottleneck?

2. Data Layout in SSTables:
   How are scalar data and vector data physically stored within the SSTables?

Are they stored in-row (scalar and vector attributes residing together in the same data block/row)?

Or are they stored in a decoupled/columnar fashion (vectors stored in a separate area/file from the scalar attributes) to optimize disk I/O during large-scale scans?

No files, tests, or entry points are named. Investigate the filtered vector-search indexing path and SSTable storage layout; done means documenting whether scalar and vector filtering or indexes are combined and how scalar and vector data are physically stored.