The single source of truth for the Khepri stream layout.

reckon-db stores events under a **structural aggregate-type subtree** (DESIGN_STREAM_NAMESPACE.md, Model C):

  USER:    [streams, <<"ride">>, <<"abc…hex">>, PaddedVersion] -> #event{}
  SYSTEM:  [streams, <<"$link">>, <<"name">>,    PaddedVersion] -> #event{}
  DCB:     [streams, <<"_dcb">>,  SeqKey]                        -> #event{}

The aggregate **type** (the user-id prefix, or the $-namespace for system streams) is a real Khepri path level rather than a leading substring of an opaque id. This makes type-scoped operations (list_streams, replay, links) navigate a subtree instead of scanning the whole store.

This module is the ONLY place that knows the 4-level shape. Every other module builds stream paths through these functions, so the layout lives in exactly one file. The opaque round-trip — a stream id written under event_path/2 must reconstruct byte-identical via stream_id_from_path/1 — is the load-bearing correctness property and is exercised directly by reckon_db_stream_path_tests.

The DCB pseudo-stream (_dcb) keeps its existing 2-level [streams, _dcb, SeqKey] shape (a flat seq-keyed log, not an aggregate); its paths are built by reckon_db_dcb_paths, but the readers here (stream_id_from_path/1) tolerate it because global reads span both depths.